Upload 144 files

.gitattributes

@@ -33,3 +33,8 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+src/rotator_library/__pycache__/client.cpython-311.pyc filter=lfs diff=lfs merge=lfs -text
+src/rotator_library/__pycache__/client.cpython-314.pyc filter=lfs diff=lfs merge=lfs -text
+src/rotator_library/__pycache__/credential_tool.cpython-311.pyc filter=lfs diff=lfs merge=lfs -text
+src/rotator_library/__pycache__/usage_manager.cpython-311.pyc filter=lfs diff=lfs merge=lfs -text
+src/rotator_library/providers/__pycache__/antigravity_provider.cpython-311.pyc filter=lfs diff=lfs merge=lfs -text

Dockerfile

@@ -0,0 +1,49 @@
+# Build stage
+FROM python:3.11-slim AS builder
+
+WORKDIR /app
+
+# Install build dependencies
+RUN apt-get update && apt-get install -y --no-install-recommends \
+    gcc \
+    && rm -rf /var/lib/apt/lists/*
+
+# Set PATH for user-installed packages in builder stage
+ENV PATH=/root/.local/bin:$PATH
+
+# Copy requirements first for better caching
+COPY requirements.txt .
+
+# Copy the local rotator_library for editable install
+COPY src/rotator_library ./src/rotator_library
+
+# Install dependencies
+RUN pip install --no-cache-dir --user -r requirements.txt
+
+# Production stage
+FROM python:3.11-slim
+
+WORKDIR /app
+
+# Copy installed packages from builder
+COPY --from=builder /root/.local /root/.local
+
+# Make sure scripts in .local are usable
+ENV PATH=/root/.local/bin:$PATH
+
+# Copy application code
+COPY src/ ./src/
+
+# Create directories for logs and oauth credentials
+RUN mkdir -p logs oauth_creds
+
+# Expose the default Hugging Face port
+EXPOSE 7860
+
+# Set environment variables
+ENV PYTHONUNBUFFERED=1
+ENV PYTHONDONTWRITEBYTECODE=1
+ENV PYTHONPATH=/app/src
+
+# Default command - runs proxy on HF's expected port
+CMD ["python", "src/proxy_app/main.py", "--host", "0.0.0.0", "--port", "7860"]

logs/failures.log

@@ -0,0 +1,10 @@
+{"timestamp": "2026-01-23T15:44:48.444258", "api_key_ending": "antigravity_oauth_1.json", "model": "antigravity/gemini-2.0-flash-exp", "attempt_number": 1, "error_type": "TransientQuotaError", "error_message": "The model returned transient 429 errors after multiple attempts. This may indicate a temporary service issue. Please try again.", "raw_response": "The model returned transient 429 errors after multiple attempts. This may indicate a temporary service issue. Please try again.", "request_headers": {"host": "localhost:8000", "user-agent": "curl/8.5.0", "accept": "*/*", "content-type": "application/json", "authorization": "Bearer sk-antigravity-proxy-123", "content-length": "114"}, "error_chain": [{"type": "TransientQuotaError", "message": "The model returned transient 429 errors after multiple attempts. This may indicate a temporary service issue. Please try again."}, {"type": "HTTPStatusError", "message": "Client error '429 Too Many Requests' for url 'https://cloudcode-pa.googleapis.com/v1internal:streamGenerateContent?alt=sse'\nFor more information check: https://developer.mozilla.org/en-US/docs/Web/HTTP/Status/429"}]}
+{"timestamp": "2026-01-23T15:45:05.739946", "api_key_ending": "antigravity_oauth_2.json", "model": "antigravity/gemini-2.0-flash-exp", "attempt_number": 1, "error_type": "TransientQuotaError", "error_message": "The model returned transient 429 errors after multiple attempts. This may indicate a temporary service issue. Please try again.", "raw_response": "The model returned transient 429 errors after multiple attempts. This may indicate a temporary service issue. Please try again.", "request_headers": {"host": "localhost:8000", "user-agent": "curl/8.5.0", "accept": "*/*", "content-type": "application/json", "authorization": "Bearer sk-antigravity-proxy-123", "content-length": "114"}, "error_chain": [{"type": "TransientQuotaError", "message": "The model returned transient 429 errors after multiple attempts. This may indicate a temporary service issue. Please try again."}, {"type": "HTTPStatusError", "message": "Client error '429 Too Many Requests' for url 'https://cloudcode-pa.googleapis.com/v1internal:streamGenerateContent?alt=sse'\nFor more information check: https://developer.mozilla.org/en-US/docs/Web/HTTP/Status/429"}]}
+{"timestamp": "2026-01-23T15:45:37.861148", "api_key_ending": "antigravity_oauth_1.json", "model": "antigravity/gemini-1.5-pro", "attempt_number": 1, "error_type": "TransientQuotaError", "error_message": "The model returned transient 429 errors after multiple attempts. This may indicate a temporary service issue. Please try again.", "raw_response": "The model returned transient 429 errors after multiple attempts. This may indicate a temporary service issue. Please try again.", "request_headers": {"host": "localhost:8000", "user-agent": "curl/8.5.0", "accept": "*/*", "content-type": "application/json", "authorization": "Bearer sk-antigravity-proxy-123", "content-length": "104"}, "error_chain": [{"type": "TransientQuotaError", "message": "The model returned transient 429 errors after multiple attempts. This may indicate a temporary service issue. Please try again."}, {"type": "HTTPStatusError", "message": "Client error '429 Too Many Requests' for url 'https://cloudcode-pa.googleapis.com/v1internal:streamGenerateContent?alt=sse'\nFor more information check: https://developer.mozilla.org/en-US/docs/Web/HTTP/Status/429"}]}
+{"timestamp": "2026-01-23T15:45:54.037894", "api_key_ending": "antigravity_oauth_2.json", "model": "antigravity/gemini-1.5-pro", "attempt_number": 1, "error_type": "TransientQuotaError", "error_message": "The model returned transient 429 errors after multiple attempts. This may indicate a temporary service issue. Please try again.", "raw_response": "The model returned transient 429 errors after multiple attempts. This may indicate a temporary service issue. Please try again.", "request_headers": {"host": "localhost:8000", "user-agent": "curl/8.5.0", "accept": "*/*", "content-type": "application/json", "authorization": "Bearer sk-antigravity-proxy-123", "content-length": "104"}, "error_chain": [{"type": "TransientQuotaError", "message": "The model returned transient 429 errors after multiple attempts. This may indicate a temporary service issue. Please try again."}, {"type": "HTTPStatusError", "message": "Client error '429 Too Many Requests' for url 'https://cloudcode-pa.googleapis.com/v1internal:streamGenerateContent?alt=sse'\nFor more information check: https://developer.mozilla.org/en-US/docs/Web/HTTP/Status/429"}]}
+{"timestamp": "2026-01-23T15:56:41.160374", "api_key_ending": "antigravity_oauth_1.json", "model": "antigravity/gemini-2.0-flash-exp", "attempt_number": 1, "error_type": "TransientQuotaError", "error_message": "The model returned transient 429 errors after multiple attempts. This may indicate a temporary service issue. Please try again.", "raw_response": "The model returned transient 429 errors after multiple attempts. This may indicate a temporary service issue. Please try again.", "request_headers": {"host": "localhost:8000", "user-agent": "curl/8.5.0", "accept": "*/*", "content-type": "application/json", "authorization": "Bearer sk-antigravity-proxy-123", "content-length": "195"}, "error_chain": [{"type": "TransientQuotaError", "message": "The model returned transient 429 errors after multiple attempts. This may indicate a temporary service issue. Please try again."}, {"type": "HTTPStatusError", "message": "Client error '429 Too Many Requests' for url 'https://cloudcode-pa.googleapis.com/v1internal:streamGenerateContent?alt=sse'\nFor more information check: https://developer.mozilla.org/en-US/docs/Web/HTTP/Status/429"}]}
+{"timestamp": "2026-01-23T15:56:58.233365", "api_key_ending": "antigravity_oauth_2.json", "model": "antigravity/gemini-2.0-flash-exp", "attempt_number": 1, "error_type": "TransientQuotaError", "error_message": "The model returned transient 429 errors after multiple attempts. This may indicate a temporary service issue. Please try again.", "raw_response": "The model returned transient 429 errors after multiple attempts. This may indicate a temporary service issue. Please try again.", "request_headers": {"host": "localhost:8000", "user-agent": "curl/8.5.0", "accept": "*/*", "content-type": "application/json", "authorization": "Bearer sk-antigravity-proxy-123", "content-length": "195"}, "error_chain": [{"type": "TransientQuotaError", "message": "The model returned transient 429 errors after multiple attempts. This may indicate a temporary service issue. Please try again."}, {"type": "HTTPStatusError", "message": "Client error '429 Too Many Requests' for url 'https://cloudcode-pa.googleapis.com/v1internal:streamGenerateContent?alt=sse'\nFor more information check: https://developer.mozilla.org/en-US/docs/Web/HTTP/Status/429"}]}
+{"timestamp": "2026-01-23T15:59:24.448647", "api_key_ending": "antigravity_oauth_2.json", "model": "antigravity/gemini-3-flash", "attempt_number": 1, "error_type": "TransientQuotaError", "error_message": "The model returned transient 429 errors after multiple attempts. This may indicate a temporary service issue. Please try again.", "raw_response": "The model returned transient 429 errors after multiple attempts. This may indicate a temporary service issue. Please try again.", "request_headers": {"host": "localhost:8000", "user-agent": "curl/8.18.0", "accept": "*/*", "content-type": "application/json", "authorization": "Bearer sk-antigravity-proxy-123", "content-length": "165"}, "error_chain": [{"type": "TransientQuotaError", "message": "The model returned transient 429 errors after multiple attempts. This may indicate a temporary service issue. Please try again."}, {"type": "HTTPStatusError", "message": "Client error '429 Too Many Requests' for url 'https://cloudcode-pa.googleapis.com/v1internal:streamGenerateContent?alt=sse'\nFor more information check: https://developer.mozilla.org/en-US/docs/Web/HTTP/Status/429"}]}
+{"timestamp": "2026-01-23T15:59:40.446459", "api_key_ending": "antigravity_oauth_1.json", "model": "antigravity/gemini-3-flash", "attempt_number": 1, "error_type": "TransientQuotaError", "error_message": "The model returned transient 429 errors after multiple attempts. This may indicate a temporary service issue. Please try again.", "raw_response": "The model returned transient 429 errors after multiple attempts. This may indicate a temporary service issue. Please try again.", "request_headers": {"host": "localhost:8000", "user-agent": "curl/8.18.0", "accept": "*/*", "content-type": "application/json", "authorization": "Bearer sk-antigravity-proxy-123", "content-length": "165"}, "error_chain": [{"type": "TransientQuotaError", "message": "The model returned transient 429 errors after multiple attempts. This may indicate a temporary service issue. Please try again."}, {"type": "HTTPStatusError", "message": "Client error '429 Too Many Requests' for url 'https://cloudcode-pa.googleapis.com/v1internal:streamGenerateContent?alt=sse'\nFor more information check: https://developer.mozilla.org/en-US/docs/Web/HTTP/Status/429"}]}
+{"timestamp": "2026-01-23T16:01:26.088546", "api_key_ending": "antigravity_oauth_2.json", "model": "antigravity/gemini-2.5-flash", "attempt_number": 1, "error_type": "TransientQuotaError", "error_message": "The model returned transient 429 errors after multiple attempts. This may indicate a temporary service issue. Please try again.", "raw_response": "The model returned transient 429 errors after multiple attempts. This may indicate a temporary service issue. Please try again.", "request_headers": {"host": "localhost:8000", "user-agent": "curl/8.18.0", "accept": "*/*", "content-type": "application/json", "authorization": "Bearer sk-antigravity-proxy-123", "content-length": "173"}, "error_chain": [{"type": "TransientQuotaError", "message": "The model returned transient 429 errors after multiple attempts. This may indicate a temporary service issue. Please try again."}, {"type": "HTTPStatusError", "message": "Client error '429 Too Many Requests' for url 'https://cloudcode-pa.googleapis.com/v1internal:streamGenerateContent?alt=sse'\nFor more information check: https://developer.mozilla.org/en-US/docs/Web/HTTP/Status/429"}]}
+{"timestamp": "2026-01-23T16:01:43.155177", "api_key_ending": "antigravity_oauth_1.json", "model": "antigravity/gemini-2.5-flash", "attempt_number": 1, "error_type": "TransientQuotaError", "error_message": "The model returned transient 429 errors after multiple attempts. This may indicate a temporary service issue. Please try again.", "raw_response": "The model returned transient 429 errors after multiple attempts. This may indicate a temporary service issue. Please try again.", "request_headers": {"host": "localhost:8000", "user-agent": "curl/8.18.0", "accept": "*/*", "content-type": "application/json", "authorization": "Bearer sk-antigravity-proxy-123", "content-length": "173"}, "error_chain": [{"type": "TransientQuotaError", "message": "The model returned transient 429 errors after multiple attempts. This may indicate a temporary service issue. Please try again."}, {"type": "HTTPStatusError", "message": "Client error '429 Too Many Requests' for url 'https://cloudcode-pa.googleapis.com/v1internal:streamGenerateContent?alt=sse'\nFor more information check: https://developer.mozilla.org/en-US/docs/Web/HTTP/Status/429"}]}

logs/proxy.log

@@ -0,0 +1,315 @@
+2026-01-23 15:44:10,808 - rotator_library - INFO - Provider 'antigravity' using rotation mode: sequential
+2026-01-23 15:44:10,809 - rotator_library - INFO - Provider 'antigravity' priority multipliers: {1: 5, 2: 3}
+2026-01-23 15:44:10,809 - rotator_library - INFO - Provider 'antigravity' sequential fallback multiplier: 2x
+2026-01-23 15:44:10,819 - rotator_library - INFO - Background token refresher started. Check interval: 600 seconds.
+2026-01-23 15:44:10,819 - root - INFO - RotatingClient initialized (EmbeddingBatcher disabled).
+2026-01-23 15:44:10,820 - rotator_library.model_info_service - INFO - ModelRegistry started (refresh every 21600s)
+2026-01-23 15:44:10,820 - root - INFO - Model info service started (fetching pricing data in background).
+2026-01-23 15:44:10,822 - rotator_library - INFO - Providers initialized: 1 providers, 2 credentials
+2026-01-23 15:44:10,823 - rotator_library - INFO -   OAuth: antigravity:2 (standard-tier:2)
+2026-01-23 15:44:10,823 - rotator_library - INFO - Started antigravity antigravity_quota_refresh (interval: 300s)
+2026-01-23 15:44:10,825 - rotator_library - INFO - antigravity: Fetching initial quota baselines for 2 credentials...
+2026-01-23 15:44:11,651 - rotator_library.model_info_service - INFO - OpenRouter: 345 models loaded
+2026-01-23 15:44:11,651 - rotator_library.model_info_service - INFO - Models.dev: 2255 models loaded
+2026-01-23 15:44:28,311 - root - INFO - 15:44 - 172.17.0.1:42688 - provider: antigravity, model: gemini-2.0-flash-exp - N/A
+2026-01-23 15:44:28,312 - rotator_library - INFO - Acquiring key for model antigravity/gemini-2.0-flash-exp. Tried keys: 0/2(2)
+2026-01-23 15:44:28,317 - rotator_library - INFO - Acquired key antigravity_oauth_1.json for model antigravity/gemini-2.0-flash-exp (tier: standard-tier, priority: 2, selection: sequential, quota: 0)
+2026-01-23 15:44:28,318 - rotator_library - INFO - Attempting call with credential antigravity_oauth_1.json (Attempt 1/2)
+2026-01-23 15:44:29,652 - rotator_library - INFO - Switching to fallback URL: https://daily-cloudcode-pa.googleapis.com/v1internal
+2026-01-23 15:44:29,652 - rotator_library - WARNING - Retrying with fallback URL: Client error '404 Not Found' for url 'https://daily-cloudcode-pa.sandbox.googleapis.com/v1internal:streamGenerateContent?alt=sse'
+For more information check: https://developer.mozilla.org/en-US/docs/Web/HTTP/Status/404
+2026-01-23 15:44:30,953 - rotator_library - INFO - Switching to fallback URL: https://cloudcode-pa.googleapis.com/v1internal
+2026-01-23 15:44:30,954 - rotator_library - WARNING - Retrying with fallback URL: Client error '404 Not Found' for url 'https://daily-cloudcode-pa.googleapis.com/v1internal:streamGenerateContent?alt=sse'
+For more information check: https://developer.mozilla.org/en-US/docs/Web/HTTP/Status/404
+2026-01-23 15:44:31,958 - rotator_library - WARNING - [Antigravity] Bare 429 from gemini-2.0-flash-exp, attempt 1/6. Retrying...
+2026-01-23 15:44:35,483 - rotator_library - WARNING - [Antigravity] Bare 429 from gemini-2.0-flash-exp, attempt 2/6. Retrying...
+2026-01-23 15:44:39,025 - rotator_library - WARNING - [Antigravity] Bare 429 from gemini-2.0-flash-exp, attempt 3/6. Retrying...
+2026-01-23 15:44:42,173 - rotator_library - WARNING - [Antigravity] Bare 429 from gemini-2.0-flash-exp, attempt 4/6. Retrying...
+2026-01-23 15:44:45,306 - rotator_library - WARNING - [Antigravity] Bare 429 from gemini-2.0-flash-exp, attempt 5/6. Retrying...
+2026-01-23 15:44:48,445 - rotator_library - ERROR - API call failed for model antigravity/gemini-2.0-flash-exp with key antigravity_oauth_1.json. Error: TransientQuotaError. See failures.log for details.
+2026-01-23 15:44:48,445 - rotator_library - WARNING - Cred antigravity_oauth_1.json server_error (HTTP 503).
+2026-01-23 15:44:48,446 - rotator_library - INFO - Provider-level error (server_error) for key antigravity_oauth_1.json with model antigravity/gemini-2.0-flash-exp. NOT incrementing failures. Cooldown: 30s
+2026-01-23 15:44:48,447 - rotator_library - INFO - Released credential antigravity_oauth_1.json from model antigravity/gemini-2.0-flash-exp (remaining concurrent: 0)
+2026-01-23 15:44:48,447 - rotator_library - INFO - Acquiring key for model antigravity/gemini-2.0-flash-exp. Tried keys: 1/1(2)
+2026-01-23 15:44:48,448 - rotator_library - INFO - Acquired key antigravity_oauth_2.json for model antigravity/gemini-2.0-flash-exp (tier: standard-tier, priority: 2, selection: sequential, quota: 0)
+2026-01-23 15:44:48,448 - rotator_library - INFO - Attempting call with credential antigravity_oauth_2.json (Attempt 1/2)
+2026-01-23 15:44:49,270 - rotator_library - WARNING - [Antigravity] Bare 429 from gemini-2.0-flash-exp, attempt 1/6. Retrying...
+2026-01-23 15:44:52,782 - rotator_library - WARNING - [Antigravity] Bare 429 from gemini-2.0-flash-exp, attempt 2/6. Retrying...
+2026-01-23 15:44:56,295 - rotator_library - WARNING - [Antigravity] Bare 429 from gemini-2.0-flash-exp, attempt 3/6. Retrying...
+2026-01-23 15:44:59,436 - rotator_library - WARNING - [Antigravity] Bare 429 from gemini-2.0-flash-exp, attempt 4/6. Retrying...
+2026-01-23 15:45:02,593 - rotator_library - WARNING - [Antigravity] Bare 429 from gemini-2.0-flash-exp, attempt 5/6. Retrying...
+2026-01-23 15:45:05,740 - rotator_library - ERROR - API call failed for model antigravity/gemini-2.0-flash-exp with key antigravity_oauth_2.json. Error: TransientQuotaError. See failures.log for details.
+2026-01-23 15:45:05,740 - rotator_library - WARNING - Cred antigravity_oauth_2.json server_error (HTTP 503).
+2026-01-23 15:45:05,740 - rotator_library - INFO - Provider-level error (server_error) for key antigravity_oauth_2.json with model antigravity/gemini-2.0-flash-exp. NOT incrementing failures. Cooldown: 30s
+2026-01-23 15:45:05,741 - rotator_library - INFO - Released credential antigravity_oauth_2.json from model antigravity/gemini-2.0-flash-exp (remaining concurrent: 0)
+2026-01-23 15:45:05,742 - rotator_library - ERROR - TIMEOUT: 2 creds tried for antigravity/gemini-2.0-flash-exp | Normal: 2 server_error
+2026-01-23 15:45:21,583 - root - INFO - 15:45 - 172.17.0.1:53018 - provider: antigravity, model: gemini-1.5-pro - N/A
+2026-01-23 15:45:21,584 - rotator_library - INFO - Acquiring key for model antigravity/gemini-1.5-pro. Tried keys: 0/2(2)
+2026-01-23 15:45:21,586 - rotator_library - INFO - Acquired key antigravity_oauth_1.json for model antigravity/gemini-1.5-pro (tier: standard-tier, priority: 2, selection: sequential, quota: 0)
+2026-01-23 15:45:21,586 - rotator_library - INFO - Attempting call with credential antigravity_oauth_1.json (Attempt 1/2)
+2026-01-23 15:45:22,152 - rotator_library - WARNING - [Antigravity] Bare 429 from gemini-1.5-pro, attempt 1/6. Retrying...
+2026-01-23 15:45:25,281 - rotator_library - WARNING - [Antigravity] Bare 429 from gemini-1.5-pro, attempt 2/6. Retrying...
+2026-01-23 15:45:28,411 - rotator_library - WARNING - [Antigravity] Bare 429 from gemini-1.5-pro, attempt 3/6. Retrying...
+2026-01-23 15:45:31,562 - rotator_library - WARNING - [Antigravity] Bare 429 from gemini-1.5-pro, attempt 4/6. Retrying...
+2026-01-23 15:45:34,708 - rotator_library - WARNING - [Antigravity] Bare 429 from gemini-1.5-pro, attempt 5/6. Retrying...
+2026-01-23 15:45:37,861 - rotator_library - ERROR - API call failed for model antigravity/gemini-1.5-pro with key antigravity_oauth_1.json. Error: TransientQuotaError. See failures.log for details.
+2026-01-23 15:45:37,862 - rotator_library - WARNING - Cred antigravity_oauth_1.json server_error (HTTP 503).
+2026-01-23 15:45:37,862 - rotator_library - INFO - Provider-level error (server_error) for key antigravity_oauth_1.json with model antigravity/gemini-1.5-pro. NOT incrementing failures. Cooldown: 30s
+2026-01-23 15:45:37,863 - rotator_library - INFO - Released credential antigravity_oauth_1.json from model antigravity/gemini-1.5-pro (remaining concurrent: 0)
+2026-01-23 15:45:37,864 - rotator_library - INFO - Acquiring key for model antigravity/gemini-1.5-pro. Tried keys: 1/1(2)
+2026-01-23 15:45:37,865 - rotator_library - INFO - Acquired key antigravity_oauth_2.json for model antigravity/gemini-1.5-pro (tier: standard-tier, priority: 2, selection: sequential, quota: 0)
+2026-01-23 15:45:37,866 - rotator_library - INFO - Attempting call with credential antigravity_oauth_2.json (Attempt 1/2)
+2026-01-23 15:45:38,007 - rotator_library - WARNING - [Antigravity] Bare 429 from gemini-1.5-pro, attempt 1/6. Retrying...
+2026-01-23 15:45:41,153 - rotator_library - WARNING - [Antigravity] Bare 429 from gemini-1.5-pro, attempt 2/6. Retrying...
+2026-01-23 15:45:44,308 - rotator_library - WARNING - [Antigravity] Bare 429 from gemini-1.5-pro, attempt 3/6. Retrying...
+2026-01-23 15:45:47,477 - rotator_library - WARNING - [Antigravity] Bare 429 from gemini-1.5-pro, attempt 4/6. Retrying...
+2026-01-23 15:45:50,630 - rotator_library - WARNING - [Antigravity] Bare 429 from gemini-1.5-pro, attempt 5/6. Retrying...
+2026-01-23 15:45:54,038 - rotator_library - ERROR - API call failed for model antigravity/gemini-1.5-pro with key antigravity_oauth_2.json. Error: TransientQuotaError. See failures.log for details.
+2026-01-23 15:45:54,038 - rotator_library - WARNING - Cred antigravity_oauth_2.json server_error (HTTP 503).
+2026-01-23 15:45:54,039 - rotator_library - INFO - Provider-level error (server_error) for key antigravity_oauth_2.json with model antigravity/gemini-1.5-pro. NOT incrementing failures. Cooldown: 30s
+2026-01-23 15:45:54,041 - rotator_library - INFO - Released credential antigravity_oauth_2.json from model antigravity/gemini-1.5-pro (remaining concurrent: 0)
+2026-01-23 15:45:54,041 - rotator_library - ERROR - TIMEOUT: 2 creds tried for antigravity/gemini-1.5-pro | Normal: 2 server_error
+2026-01-23 15:56:23,674 - root - INFO - 15:56 - 172.17.0.1:41590 - provider: antigravity, model: gemini-2.0-flash-exp - N/A
+2026-01-23 15:56:23,675 - rotator_library - INFO - Acquiring key for model antigravity/gemini-2.0-flash-exp. Tried keys: 0/2(2)
+2026-01-23 15:56:23,677 - rotator_library - INFO - Acquired key antigravity_oauth_1.json for model antigravity/gemini-2.0-flash-exp (tier: standard-tier, priority: 2, selection: sequential, quota: 1)
+2026-01-23 15:56:23,677 - rotator_library - INFO - Attempting call with credential antigravity_oauth_1.json (Attempt 1/2)
+2026-01-23 15:56:24,692 - rotator_library - WARNING - [Antigravity] Bare 429 from gemini-2.0-flash-exp, attempt 1/6. Retrying...
+2026-01-23 15:56:28,214 - rotator_library - WARNING - [Antigravity] Bare 429 from gemini-2.0-flash-exp, attempt 2/6. Retrying...
+2026-01-23 15:56:31,724 - rotator_library - WARNING - [Antigravity] Bare 429 from gemini-2.0-flash-exp, attempt 3/6. Retrying...
+2026-01-23 15:56:34,859 - rotator_library - WARNING - [Antigravity] Bare 429 from gemini-2.0-flash-exp, attempt 4/6. Retrying...
+2026-01-23 15:56:38,005 - rotator_library - WARNING - [Antigravity] Bare 429 from gemini-2.0-flash-exp, attempt 5/6. Retrying...
+2026-01-23 15:56:41,161 - rotator_library - ERROR - API call failed for model antigravity/gemini-2.0-flash-exp with key antigravity_oauth_1.json. Error: TransientQuotaError. See failures.log for details.
+2026-01-23 15:56:41,161 - rotator_library - WARNING - Cred antigravity_oauth_1.json server_error (HTTP 503).
+2026-01-23 15:56:41,162 - rotator_library - INFO - Provider-level error (server_error) for key antigravity_oauth_1.json with model antigravity/gemini-2.0-flash-exp. NOT incrementing failures. Cooldown: 30s
+2026-01-23 15:56:41,163 - rotator_library - INFO - Released credential antigravity_oauth_1.json from model antigravity/gemini-2.0-flash-exp (remaining concurrent: 0)
+2026-01-23 15:56:41,163 - rotator_library - INFO - Acquiring key for model antigravity/gemini-2.0-flash-exp. Tried keys: 1/1(2)
+2026-01-23 15:56:41,164 - rotator_library - INFO - Acquired key antigravity_oauth_2.json for model antigravity/gemini-2.0-flash-exp (tier: standard-tier, priority: 2, selection: sequential, quota: 1)
+2026-01-23 15:56:41,164 - rotator_library - INFO - Attempting call with credential antigravity_oauth_2.json (Attempt 1/2)
+2026-01-23 15:56:41,758 - rotator_library - WARNING - [Antigravity] Bare 429 from gemini-2.0-flash-exp, attempt 1/6. Retrying...
+2026-01-23 15:56:45,275 - rotator_library - WARNING - [Antigravity] Bare 429 from gemini-2.0-flash-exp, attempt 2/6. Retrying...
+2026-01-23 15:56:48,810 - rotator_library - WARNING - [Antigravity] Bare 429 from gemini-2.0-flash-exp, attempt 3/6. Retrying...
+2026-01-23 15:56:51,963 - rotator_library - WARNING - [Antigravity] Bare 429 from gemini-2.0-flash-exp, attempt 4/6. Retrying...
+2026-01-23 15:56:55,088 - rotator_library - WARNING - [Antigravity] Bare 429 from gemini-2.0-flash-exp, attempt 5/6. Retrying...
+2026-01-23 15:56:58,233 - rotator_library - ERROR - API call failed for model antigravity/gemini-2.0-flash-exp with key antigravity_oauth_2.json. Error: TransientQuotaError. See failures.log for details.
+2026-01-23 15:56:58,234 - rotator_library - WARNING - Cred antigravity_oauth_2.json server_error (HTTP 503).
+2026-01-23 15:56:58,234 - rotator_library - INFO - Provider-level error (server_error) for key antigravity_oauth_2.json with model antigravity/gemini-2.0-flash-exp. NOT incrementing failures. Cooldown: 30s
+2026-01-23 15:56:58,236 - rotator_library - INFO - Released credential antigravity_oauth_2.json from model antigravity/gemini-2.0-flash-exp (remaining concurrent: 0)
+2026-01-23 15:56:58,236 - rotator_library - ERROR - TIMEOUT: 2 creds tried for antigravity/gemini-2.0-flash-exp | Normal: 2 server_error
+2026-01-23 15:58:31,358 - rotator_library - INFO - Getting all available models...
+2026-01-23 15:58:31,358 - rotator_library - INFO - Getting available models for provider: antigravity
+2026-01-23 15:58:31,359 - rotator_library - INFO - Got 6 models for provider: antigravity
+2026-01-23 15:58:31,360 - rotator_library - INFO - Finished getting all available models.
+2026-01-23 15:59:07,976 - root - INFO - 15:59 - 172.17.0.1:41760 - provider: antigravity, model: gemini-3-flash - N/A
+2026-01-23 15:59:07,976 - rotator_library - INFO - Acquiring key for model antigravity/gemini-3-flash. Tried keys: 0/2(2)
+2026-01-23 15:59:07,978 - rotator_library - INFO - Acquired key antigravity_oauth_2.json for model antigravity/gemini-3-flash (tier: standard-tier, priority: 2, selection: sequential, quota: 79/400 [80%])
+2026-01-23 15:59:07,978 - rotator_library - INFO - Attempting call with credential antigravity_oauth_2.json (Attempt 1/2)
+2026-01-23 15:59:08,623 - rotator_library - WARNING - [Antigravity] Bare 429 from gemini-3-flash, attempt 1/6. Retrying...
+2026-01-23 15:59:11,803 - rotator_library - WARNING - [Antigravity] Bare 429 from gemini-3-flash, attempt 2/6. Retrying...
+2026-01-23 15:59:15,012 - rotator_library - WARNING - [Antigravity] Bare 429 from gemini-3-flash, attempt 3/6. Retrying...
+2026-01-23 15:59:18,150 - rotator_library - WARNING - [Antigravity] Bare 429 from gemini-3-flash, attempt 4/6. Retrying...
+2026-01-23 15:59:21,293 - rotator_library - WARNING - [Antigravity] Bare 429 from gemini-3-flash, attempt 5/6. Retrying...
+2026-01-23 15:59:24,449 - rotator_library - ERROR - API call failed for model antigravity/gemini-3-flash with key antigravity_oauth_2.json. Error: TransientQuotaError. See failures.log for details.
+2026-01-23 15:59:24,449 - rotator_library - WARNING - Cred antigravity_oauth_2.json server_error (HTTP 503).
+2026-01-23 15:59:24,450 - rotator_library - INFO - Provider-level error (server_error) for key antigravity_oauth_2.json with model antigravity/gemini-3-flash. NOT incrementing failures. Cooldown: 30s
+2026-01-23 15:59:24,451 - rotator_library - INFO - Released credential antigravity_oauth_2.json from model antigravity/gemini-3-flash (remaining concurrent: 0)
+2026-01-23 15:59:24,452 - rotator_library - INFO - Acquiring key for model antigravity/gemini-3-flash. Tried keys: 1/1(2)
+2026-01-23 15:59:24,453 - rotator_library - INFO - Acquired key antigravity_oauth_1.json for model antigravity/gemini-3-flash (tier: standard-tier, priority: 2, selection: sequential, quota: 0/400 [100%])
+2026-01-23 15:59:24,453 - rotator_library - INFO - Attempting call with credential antigravity_oauth_1.json (Attempt 1/2)
+2026-01-23 15:59:24,658 - rotator_library - WARNING - [Antigravity] Bare 429 from gemini-3-flash, attempt 1/6. Retrying...
+2026-01-23 15:59:27,819 - rotator_library - WARNING - [Antigravity] Bare 429 from gemini-3-flash, attempt 2/6. Retrying...
+2026-01-23 15:59:31,009 - rotator_library - WARNING - [Antigravity] Bare 429 from gemini-3-flash, attempt 3/6. Retrying...
+2026-01-23 15:59:34,150 - rotator_library - WARNING - [Antigravity] Bare 429 from gemini-3-flash, attempt 4/6. Retrying...
+2026-01-23 15:59:37,303 - rotator_library - WARNING - [Antigravity] Bare 429 from gemini-3-flash, attempt 5/6. Retrying...
+2026-01-23 15:59:40,447 - rotator_library - ERROR - API call failed for model antigravity/gemini-3-flash with key antigravity_oauth_1.json. Error: TransientQuotaError. See failures.log for details.
+2026-01-23 15:59:40,447 - rotator_library - WARNING - Cred antigravity_oauth_1.json server_error (HTTP 503).
+2026-01-23 15:59:40,448 - rotator_library - INFO - Provider-level error (server_error) for key antigravity_oauth_1.json with model antigravity/gemini-3-flash. NOT incrementing failures. Cooldown: 30s
+2026-01-23 15:59:40,450 - rotator_library - INFO - Released credential antigravity_oauth_1.json from model antigravity/gemini-3-flash (remaining concurrent: 0)
+2026-01-23 15:59:40,450 - rotator_library - ERROR - TIMEOUT: 2 creds tried for antigravity/gemini-3-flash | Normal: 2 server_error
+2026-01-23 16:00:10,954 - rotator_library - INFO - Getting all available models...
+2026-01-23 16:00:10,954 - rotator_library - INFO - Getting available models for provider: antigravity
+2026-01-23 16:00:10,955 - rotator_library - INFO - Finished getting all available models.
+2026-01-23 16:00:21,200 - rotator_library - INFO - Getting all available models...
+2026-01-23 16:00:21,201 - rotator_library - INFO - Getting available models for provider: antigravity
+2026-01-23 16:00:21,201 - rotator_library - INFO - Finished getting all available models.
+2026-01-23 16:01:08,173 - root - INFO - 16:01 - 172.17.0.1:39582 - provider: antigravity, model: gemini-2.5-flash - N/A
+2026-01-23 16:01:08,174 - rotator_library - INFO - Acquiring key for model antigravity/gemini-2.5-flash. Tried keys: 0/2(2)
+2026-01-23 16:01:08,175 - rotator_library - INFO - Acquired key antigravity_oauth_2.json for model antigravity/gemini-2.5-flash (tier: standard-tier, priority: 2, selection: sequential, quota: 599/3000 [80%])
+2026-01-23 16:01:08,176 - rotator_library - INFO - Attempting call with credential antigravity_oauth_2.json (Attempt 1/2)
+2026-01-23 16:01:08,892 - rotator_library - WARNING - [Antigravity] Bare 429 from gemini-2.5-flash, attempt 1/6. Retrying...
+2026-01-23 16:01:12,675 - rotator_library - WARNING - [Antigravity] Bare 429 from gemini-2.5-flash, attempt 2/6. Retrying...
+2026-01-23 16:01:15,849 - rotator_library - WARNING - [Antigravity] Bare 429 from gemini-2.5-flash, attempt 3/6. Retrying...
+2026-01-23 16:01:19,004 - rotator_library - WARNING - [Antigravity] Bare 429 from gemini-2.5-flash, attempt 4/6. Retrying...
+2026-01-23 16:01:22,584 - rotator_library - WARNING - [Antigravity] Bare 429 from gemini-2.5-flash, attempt 5/6. Retrying...
+2026-01-23 16:01:26,089 - rotator_library - ERROR - API call failed for model antigravity/gemini-2.5-flash with key antigravity_oauth_2.json. Error: TransientQuotaError. See failures.log for details.
+2026-01-23 16:01:26,089 - rotator_library - WARNING - Cred antigravity_oauth_2.json server_error (HTTP 503).
+2026-01-23 16:01:26,089 - rotator_library - INFO - Provider-level error (server_error) for key antigravity_oauth_2.json with model antigravity/gemini-2.5-flash. NOT incrementing failures. Cooldown: 30s
+2026-01-23 16:01:26,090 - rotator_library - INFO - Released credential antigravity_oauth_2.json from model antigravity/gemini-2.5-flash (remaining concurrent: 0)
+2026-01-23 16:01:26,091 - rotator_library - INFO - Acquiring key for model antigravity/gemini-2.5-flash. Tried keys: 1/1(2)
+2026-01-23 16:01:26,091 - rotator_library - INFO - Acquired key antigravity_oauth_1.json for model antigravity/gemini-2.5-flash (tier: standard-tier, priority: 2, selection: sequential, quota: 0/3000 [100%])
+2026-01-23 16:01:26,092 - rotator_library - INFO - Attempting call with credential antigravity_oauth_1.json (Attempt 1/2)
+2026-01-23 16:01:26,606 - rotator_library - WARNING - [Antigravity] Bare 429 from gemini-2.5-flash, attempt 1/6. Retrying...
+2026-01-23 16:01:30,115 - rotator_library - WARNING - [Antigravity] Bare 429 from gemini-2.5-flash, attempt 2/6. Retrying...
+2026-01-23 16:01:33,311 - rotator_library - WARNING - [Antigravity] Bare 429 from gemini-2.5-flash, attempt 3/6. Retrying...
+2026-01-23 16:01:36,840 - rotator_library - WARNING - [Antigravity] Bare 429 from gemini-2.5-flash, attempt 4/6. Retrying...
+2026-01-23 16:01:39,989 - rotator_library - WARNING - [Antigravity] Bare 429 from gemini-2.5-flash, attempt 5/6. Retrying...
+2026-01-23 16:01:43,155 - rotator_library - ERROR - API call failed for model antigravity/gemini-2.5-flash with key antigravity_oauth_1.json. Error: TransientQuotaError. See failures.log for details.
+2026-01-23 16:01:43,156 - rotator_library - WARNING - Cred antigravity_oauth_1.json server_error (HTTP 503).
+2026-01-23 16:01:43,156 - rotator_library - INFO - Provider-level error (server_error) for key antigravity_oauth_1.json with model antigravity/gemini-2.5-flash. NOT incrementing failures. Cooldown: 30s
+2026-01-23 16:01:43,157 - rotator_library - INFO - Released credential antigravity_oauth_1.json from model antigravity/gemini-2.5-flash (remaining concurrent: 0)
+2026-01-23 16:01:43,157 - rotator_library - ERROR - TIMEOUT: 2 creds tried for antigravity/gemini-2.5-flash | Normal: 2 server_error
+2026-01-23 16:02:46,116 - rotator_library - INFO - Background token refresher stopped.
+2026-01-23 16:02:46,116 - rotator_library.model_info_service - INFO - ModelRegistry stopped
+2026-01-23 16:02:46,116 - root - INFO - RotatingClient closed.
+2026-01-23 16:03:05,733 - rotator_library - INFO - Provider 'antigravity' using rotation mode: sequential
+2026-01-23 16:03:05,734 - rotator_library - INFO - Provider 'antigravity' priority multipliers: {1: 5, 2: 3}
+2026-01-23 16:03:05,734 - rotator_library - INFO - Provider 'antigravity' sequential fallback multiplier: 2x
+2026-01-23 16:03:05,745 - rotator_library - INFO - Background token refresher started. Check interval: 600 seconds.
+2026-01-23 16:03:05,745 - root - INFO - RotatingClient initialized (EmbeddingBatcher disabled).
+2026-01-23 16:03:05,745 - rotator_library.model_info_service - INFO - ModelRegistry started (refresh every 21600s)
+2026-01-23 16:03:05,746 - root - INFO - Model info service started (fetching pricing data in background).
+2026-01-23 16:03:05,748 - rotator_library - INFO - Providers initialized: 1 providers, 2 credentials
+2026-01-23 16:03:05,748 - rotator_library - INFO -   OAuth: antigravity:2 (standard-tier:2)
+2026-01-23 16:03:05,748 - rotator_library - INFO - Started antigravity antigravity_quota_refresh (interval: 300s)
+2026-01-23 16:03:05,751 - rotator_library - INFO - antigravity: Fetching initial quota baselines for 2 credentials...
+2026-01-23 16:03:06,489 - rotator_library.model_info_service - INFO - OpenRouter: 345 models loaded
+2026-01-23 16:03:06,489 - rotator_library.model_info_service - INFO - Models.dev: 2255 models loaded
+2026-01-23 16:04:06,556 - rotator_library - INFO - Getting all available models...
+2026-01-23 16:04:06,560 - rotator_library - INFO - Getting available models for provider: antigravity
+2026-01-23 16:04:06,564 - rotator_library - INFO - Got 6 models for provider: antigravity
+2026-01-23 16:04:06,565 - rotator_library - INFO - Finished getting all available models.
+2026-01-23 16:04:22,953 - root - INFO - 16:04 - 172.17.0.1:57468 - provider: antigravity, model: gemini-3-flash - N/A
+2026-01-23 16:04:22,955 - rotator_library - INFO - Acquiring key for model antigravity/gemini-3-flash. Tried keys: 0/2(2)
+2026-01-23 16:04:22,958 - rotator_library - INFO - Acquired key antigravity_oauth_2.json for model antigravity/gemini-3-flash (tier: standard-tier, priority: 2, selection: sequential, quota: 80/400 [80%])
+2026-01-23 16:04:22,958 - rotator_library - INFO - Attempting call with credential antigravity_oauth_2.json (Attempt 1/2)
+2026-01-23 16:04:25,936 - rotator_library - INFO - Started 5.0h window for model antigravity/gemini-3-flash on antigravity_oauth_2.json
+2026-01-23 16:04:25,936 - rotator_library - INFO - Recorded usage from response object for key antigravity_oauth_2.json
+2026-01-23 16:04:25,938 - rotator_library - INFO - Released credential antigravity_oauth_2.json from model antigravity/gemini-3-flash (remaining concurrent: 0)
+2026-01-23 16:12:00,819 - rotator_library - INFO - Background token refresher stopped.
+2026-01-23 16:12:00,820 - rotator_library.model_info_service - INFO - ModelRegistry stopped
+2026-01-23 16:12:00,820 - root - INFO - RotatingClient closed.
+2026-01-23 16:12:06,391 - rotator_library - INFO - Provider 'antigravity' using rotation mode: sequential
+2026-01-23 16:12:06,392 - rotator_library - INFO - Provider 'antigravity' priority multipliers: {1: 5, 2: 3}
+2026-01-23 16:12:06,392 - rotator_library - INFO - Provider 'antigravity' sequential fallback multiplier: 2x
+2026-01-23 16:12:06,402 - rotator_library - INFO - Background token refresher started. Check interval: 600 seconds.
+2026-01-23 16:12:06,402 - root - INFO - RotatingClient initialized (EmbeddingBatcher disabled).
+2026-01-23 16:12:06,403 - rotator_library.model_info_service - INFO - ModelRegistry started (refresh every 21600s)
+2026-01-23 16:12:06,403 - root - INFO - Model info service started (fetching pricing data in background).
+2026-01-23 16:12:06,404 - rotator_library - INFO - Providers initialized: 1 providers, 1 credentials
+2026-01-23 16:12:06,405 - rotator_library - INFO -   OAuth: antigravity:1 (standard-tier:1)
+2026-01-23 16:12:06,405 - rotator_library - INFO - Started antigravity antigravity_quota_refresh (interval: 300s)
+2026-01-23 16:12:06,407 - rotator_library - INFO - antigravity: Fetching initial quota baselines for 1 credentials...
+2026-01-23 16:12:07,541 - rotator_library.model_info_service - INFO - OpenRouter: 345 models loaded
+2026-01-23 16:12:07,541 - rotator_library.model_info_service - INFO - Models.dev: 2255 models loaded
+2026-01-23 16:38:58,754 - root - INFO - 16:38 - 172.17.0.1:50620 - provider: antigravity, model: gemini-3-flash - N/A
+2026-01-23 16:38:58,755 - rotator_library - INFO - Acquiring key for model antigravity/gemini-3-flash. Tried keys: 0/1(1)
+2026-01-23 16:38:58,757 - rotator_library - INFO - Acquired key antigravity_oauth_1.json for model antigravity/gemini-3-flash (tier: standard-tier, priority: 2, selection: sequential, quota: 1/400 [99%])
+2026-01-23 16:38:58,758 - rotator_library - INFO - Attempting call with credential antigravity_oauth_1.json (Attempt 1/2)
+2026-01-23 16:39:02,593 - rotator_library - INFO - Started 5.0h window for model antigravity/gemini-3-flash on antigravity_oauth_1.json
+2026-01-23 16:39:02,593 - rotator_library - INFO - Recorded usage from response object for key antigravity_oauth_1.json
+2026-01-23 16:39:02,595 - rotator_library - INFO - Released credential antigravity_oauth_1.json from model antigravity/gemini-3-flash (remaining concurrent: 0)
+2026-01-24 11:18:44,039 - rotator_library - INFO - Getting all available models...
+2026-01-24 11:18:44,058 - rotator_library - INFO - Getting available models for provider: antigravity
+2026-01-24 11:18:44,059 - rotator_library - INFO - Got 6 models for provider: antigravity
+2026-01-24 11:18:44,060 - rotator_library - INFO - Finished getting all available models.
+2026-01-24 11:23:14,658 - rotator_library - INFO - Getting all available models...
+2026-01-24 11:23:14,659 - rotator_library - INFO - Getting available models for provider: antigravity
+2026-01-24 11:23:14,659 - rotator_library - INFO - Finished getting all available models.
+2026-01-24 11:32:50,410 - root - INFO - 11:32 - 172.17.0.1:45252 - provider: antigravity, model: gemini-3-flash - N/A
+2026-01-24 11:32:50,411 - rotator_library - INFO - Acquiring credential for model antigravity/gemini-3-flash. Tried credentials: 0/1(1)
+2026-01-24 11:32:50,413 - rotator_library - INFO - Reset model group 'g3-flash' (1 models) for antigravity_oauth_1.json
+2026-01-24 11:32:50,524 - rotator_library - INFO - Reset model group 'g3-flash' (1 models) for antigravity_oauth_2.json
+2026-01-24 11:32:50,817 - rotator_library - INFO - Acquired key antigravity_oauth_1.json for model antigravity/gemini-3-flash (tier: standard-tier, priority: 2, selection: sequential, quota: 0/400 [100%])
+2026-01-24 11:32:50,881 - rotator_library - INFO - Attempting stream with credential antigravity_oauth_1.json (Attempt 1/2)
+2026-01-24 11:32:51,096 - rotator_library - INFO - Stream connection established for credential antigravity_oauth_1.json. Processing response.
+2026-01-24 11:32:51,590 - root - INFO - 11:32 - 172.17.0.1:45266 - provider: antigravity, model: gemini-3-flash - N/A
+2026-01-24 11:32:51,597 - rotator_library - INFO - Acquiring credential for model antigravity/gemini-3-flash. Tried credentials: 0/1(1)
+2026-01-24 11:32:51,598 - rotator_library - INFO - Acquired key antigravity_oauth_1.json for model antigravity/gemini-3-flash (tier: standard-tier, priority: 2, selection: sequential, concurrent: 2/3, quota: 0/400 [100%])
+2026-01-24 11:32:51,599 - rotator_library - INFO - Attempting stream with credential antigravity_oauth_1.json (Attempt 1/2)
+2026-01-24 11:32:51,599 - rotator_library - INFO - Stream connection established for credential antigravity_oauth_1.json. Processing response.
+2026-01-24 11:32:51,607 - root - INFO - 11:32 - 172.17.0.1:45284 - provider: antigravity, model: claude-opus-4.5 - N/A
+2026-01-24 11:32:51,608 - rotator_library - INFO - Acquiring credential for model antigravity/claude-opus-4.5. Tried credentials: 0/1(1)
+2026-01-24 11:32:51,609 - rotator_library - INFO - Acquired key antigravity_oauth_1.json for model antigravity/claude-opus-4.5 (tier: standard-tier, priority: 2, selection: sequential, concurrent: 1/3, quota: 0/150 [100%])
+2026-01-24 11:32:51,610 - rotator_library - INFO - Attempting stream with credential antigravity_oauth_1.json (Attempt 1/2)
+2026-01-24 11:32:51,614 - rotator_library - INFO - Stream connection established for credential antigravity_oauth_1.json. Processing response.
+2026-01-24 11:32:54,551 - rotator_library - INFO - Started 5.0h window for model antigravity/gemini-3-flash on antigravity_oauth_1.json
+2026-01-24 11:32:54,552 - rotator_library - INFO - Recorded usage from response object for key antigravity_oauth_1.json
+2026-01-24 11:32:54,554 - rotator_library - INFO - Released credential antigravity_oauth_1.json from model antigravity/gemini-3-flash (remaining concurrent: 1)
+2026-01-24 11:32:54,554 - rotator_library - INFO - STREAM FINISHED and lock released for credential antigravity_oauth_1.json.
+2026-01-24 11:32:56,228 - rotator_library - INFO - Recorded usage from response object for key antigravity_oauth_1.json
+2026-01-24 11:32:56,229 - rotator_library - INFO - Released credential antigravity_oauth_1.json from model antigravity/gemini-3-flash (remaining concurrent: 0)
+2026-01-24 11:32:56,230 - rotator_library - INFO - STREAM FINISHED and lock released for credential antigravity_oauth_1.json.
+2026-01-24 11:32:57,435 - rotator_library - INFO - Started 5.0h window for model antigravity/claude-opus-4.5 on antigravity_oauth_1.json
+2026-01-24 11:32:57,435 - rotator_library - INFO - Recorded usage from response object for key antigravity_oauth_1.json
+2026-01-24 11:32:57,437 - rotator_library - INFO - Released credential antigravity_oauth_1.json from model antigravity/claude-opus-4.5 (remaining concurrent: 0)
+2026-01-24 11:32:57,437 - rotator_library - INFO - STREAM FINISHED and lock released for credential antigravity_oauth_1.json.
+2026-01-24 11:34:30,522 - root - INFO - 11:34 - 172.17.0.1:40506 - provider: antigravity, model: claude-opus-4.5 - N/A
+2026-01-24 11:34:30,524 - rotator_library - INFO - Acquiring credential for model antigravity/claude-opus-4.5. Tried credentials: 0/1(1)
+2026-01-24 11:34:30,526 - rotator_library - INFO - Acquired key antigravity_oauth_1.json for model antigravity/claude-opus-4.5 (tier: standard-tier, priority: 2, selection: sequential, quota: 1/150 [99%])
+2026-01-24 11:34:30,526 - rotator_library - INFO - Attempting stream with credential antigravity_oauth_1.json (Attempt 1/2)
+2026-01-24 11:34:30,645 - rotator_library - INFO - Stream connection established for credential antigravity_oauth_1.json. Processing response.
+2026-01-24 11:34:30,651 - root - INFO - 11:34 - 172.17.0.1:40538 - provider: antigravity, model: gemini-3-flash - N/A
+2026-01-24 11:34:30,652 - rotator_library - INFO - Acquiring credential for model antigravity/gemini-3-flash. Tried credentials: 0/1(1)
+2026-01-24 11:34:30,653 - rotator_library - INFO - Acquired key antigravity_oauth_1.json for model antigravity/gemini-3-flash (tier: standard-tier, priority: 2, selection: sequential, concurrent: 1/3, quota: 2/400 [99%])
+2026-01-24 11:34:30,654 - rotator_library - INFO - Attempting stream with credential antigravity_oauth_1.json (Attempt 1/2)
+2026-01-24 11:34:30,655 - rotator_library - INFO - Stream connection established for credential antigravity_oauth_1.json. Processing response.
+2026-01-24 11:34:33,407 - rotator_library - INFO - Recorded usage from response object for key antigravity_oauth_1.json
+2026-01-24 11:34:33,409 - rotator_library - INFO - Released credential antigravity_oauth_1.json from model antigravity/gemini-3-flash (remaining concurrent: 0)
+2026-01-24 11:34:33,410 - rotator_library - INFO - STREAM FINISHED and lock released for credential antigravity_oauth_1.json.
+2026-01-24 11:34:36,075 - rotator_library - INFO - Recorded usage from response object for key antigravity_oauth_1.json
+2026-01-24 11:34:36,077 - rotator_library - INFO - Released credential antigravity_oauth_1.json from model antigravity/claude-opus-4.5 (remaining concurrent: 0)
+2026-01-24 11:34:36,077 - rotator_library - INFO - STREAM FINISHED and lock released for credential antigravity_oauth_1.json.
+2026-01-24 11:34:38,581 - root - INFO - 11:34 - 172.17.0.1:40506 - provider: antigravity, model: claude-opus-4.5 - N/A
+2026-01-24 11:34:38,582 - rotator_library - INFO - Acquiring credential for model antigravity/claude-opus-4.5. Tried credentials: 0/1(1)
+2026-01-24 11:34:38,583 - rotator_library - INFO - Acquired key antigravity_oauth_1.json for model antigravity/claude-opus-4.5 (tier: standard-tier, priority: 2, selection: sequential, quota: 2/150 [98%])
+2026-01-24 11:34:38,584 - rotator_library - INFO - Attempting stream with credential antigravity_oauth_1.json (Attempt 1/2)
+2026-01-24 11:34:38,585 - rotator_library - INFO - [Thinking Sanitization] Closing tool loop - turn has no thinking at start
+2026-01-24 11:34:38,585 - rotator_library - INFO - [Thinking Sanitization] Closed tool loop with synthetic messages. Model: '[Tool execution completed.]', User: '[Continue]'. Claude will now start a fresh turn with thinking enabled.
+2026-01-24 11:34:38,589 - rotator_library - INFO - Stream connection established for credential antigravity_oauth_1.json. Processing response.
+2026-01-24 11:34:44,008 - rotator_library - INFO - Recorded usage from response object for key antigravity_oauth_1.json
+2026-01-24 11:34:44,010 - rotator_library - INFO - Released credential antigravity_oauth_1.json from model antigravity/claude-opus-4.5 (remaining concurrent: 0)
+2026-01-24 11:34:44,010 - rotator_library - INFO - STREAM FINISHED and lock released for credential antigravity_oauth_1.json.
+2026-01-24 11:38:54,076 - root - INFO - 11:38 - 172.17.0.1:37082 - provider: antigravity, model: claude-opus-4.5 - N/A
+2026-01-24 11:38:54,076 - rotator_library - INFO - Acquiring credential for model antigravity/claude-opus-4.5. Tried credentials: 0/1(1)
+2026-01-24 11:38:54,077 - rotator_library - INFO - Acquired key antigravity_oauth_1.json for model antigravity/claude-opus-4.5 (tier: standard-tier, priority: 2, selection: sequential, quota: 3/150 [98%])
+2026-01-24 11:38:54,081 - rotator_library - INFO - Attempting stream with credential antigravity_oauth_1.json (Attempt 1/2)
+2026-01-24 11:38:54,091 - rotator_library - INFO - Stream connection established for credential antigravity_oauth_1.json. Processing response.
+2026-01-24 11:38:54,149 - root - INFO - 11:38 - 172.17.0.1:37094 - provider: antigravity, model: gemini-3-flash - N/A
+2026-01-24 11:38:54,150 - rotator_library - INFO - Acquiring credential for model antigravity/gemini-3-flash. Tried credentials: 0/1(1)
+2026-01-24 11:38:54,152 - rotator_library - INFO - Acquired key antigravity_oauth_1.json for model antigravity/gemini-3-flash (tier: standard-tier, priority: 2, selection: sequential, concurrent: 1/3, quota: 3/400 [99%])
+2026-01-24 11:38:54,152 - rotator_library - INFO - Attempting stream with credential antigravity_oauth_1.json (Attempt 1/2)
+2026-01-24 11:38:54,153 - rotator_library - INFO - Stream connection established for credential antigravity_oauth_1.json. Processing response.
+2026-01-24 11:38:56,999 - rotator_library - INFO - Recorded usage from response object for key antigravity_oauth_1.json
+2026-01-24 11:38:57,001 - rotator_library - INFO - Released credential antigravity_oauth_1.json from model antigravity/gemini-3-flash (remaining concurrent: 0)
+2026-01-24 11:38:57,002 - rotator_library - INFO - STREAM FINISHED and lock released for credential antigravity_oauth_1.json.
+2026-01-24 11:38:58,817 - rotator_library - INFO - Recorded usage from response object for key antigravity_oauth_1.json
+2026-01-24 11:38:58,819 - rotator_library - INFO - Released credential antigravity_oauth_1.json from model antigravity/claude-opus-4.5 (remaining concurrent: 0)
+2026-01-24 11:38:58,820 - rotator_library - INFO - STREAM FINISHED and lock released for credential antigravity_oauth_1.json.
+2026-01-24 11:39:23,428 - root - INFO - 11:39 - 172.17.0.1:50878 - provider: antigravity, model: gemini-3-flash - N/A
+2026-01-24 11:39:23,428 - rotator_library - INFO - Acquiring credential for model antigravity/gemini-3-flash. Tried credentials: 0/1(1)
+2026-01-24 11:39:23,430 - rotator_library - INFO - Acquired key antigravity_oauth_1.json for model antigravity/gemini-3-flash (tier: standard-tier, priority: 2, selection: sequential, quota: 4/400 [99%])
+2026-01-24 11:39:23,430 - rotator_library - INFO - Attempting stream with credential antigravity_oauth_1.json (Attempt 1/2)
+2026-01-24 11:39:23,443 - rotator_library - INFO - Stream connection established for credential antigravity_oauth_1.json. Processing response.
+2026-01-24 11:39:23,450 - root - INFO - 11:39 - 172.17.0.1:50886 - provider: antigravity, model: gemini-3-flash - N/A
+2026-01-24 11:39:23,451 - rotator_library - INFO - Acquiring credential for model antigravity/gemini-3-flash. Tried credentials: 0/1(1)
+2026-01-24 11:39:23,453 - rotator_library - INFO - Acquired key antigravity_oauth_1.json for model antigravity/gemini-3-flash (tier: standard-tier, priority: 2, selection: sequential, concurrent: 2/3, quota: 4/400 [99%])
+2026-01-24 11:39:23,453 - rotator_library - INFO - Attempting stream with credential antigravity_oauth_1.json (Attempt 1/2)
+2026-01-24 11:39:23,454 - rotator_library - INFO - Stream connection established for credential antigravity_oauth_1.json. Processing response.
+2026-01-24 11:39:25,862 - rotator_library - INFO - Recorded usage from response object for key antigravity_oauth_1.json
+2026-01-24 11:39:25,865 - rotator_library - INFO - Released credential antigravity_oauth_1.json from model antigravity/gemini-3-flash (remaining concurrent: 1)
+2026-01-24 11:39:25,865 - rotator_library - INFO - STREAM FINISHED and lock released for credential antigravity_oauth_1.json.
+2026-01-24 11:39:27,283 - rotator_library - INFO - Recorded usage from response object for key antigravity_oauth_1.json
+2026-01-24 11:39:27,284 - rotator_library - INFO - Released credential antigravity_oauth_1.json from model antigravity/gemini-3-flash (remaining concurrent: 0)
+2026-01-24 11:39:27,285 - rotator_library - INFO - STREAM FINISHED and lock released for credential antigravity_oauth_1.json.
+2026-01-24 11:45:15,687 - rotator_library.model_info_service - INFO - Scheduled registry refresh...
+2026-01-24 11:45:17,015 - rotator_library.model_info_service - INFO - OpenRouter: 345 models loaded
+2026-01-24 11:45:17,018 - rotator_library.model_info_service - INFO - Models.dev: 2253 models loaded
+2026-01-24 11:45:17,035 - rotator_library.model_info_service - INFO - Registry refresh complete
+2026-01-24 17:05:30,072 - rotator_library - WARNING - Network error during refresh: [Errno -3] Temporary failure in name resolution, retry 1/3 in 1s
+2026-01-24 17:05:34,907 - rotator_library - WARNING - Refresh timeout (15s) for 'antigravity_oauth_1.json'
+2026-01-24 17:05:34,908 - rotator_library - WARNING - Refresh failed for 'antigravity_oauth_1.json' (timeout). Retry 1/3, back of queue.
+2026-01-24 17:06:14,940 - rotator_library - WARNING - Network error during refresh: [Errno -3] Temporary failure in name resolution, retry 1/3 in 1s
+2026-01-24 17:06:19,910 - rotator_library - WARNING - Refresh timeout (15s) for 'antigravity_oauth_1.json'
+2026-01-24 17:06:19,910 - rotator_library - WARNING - Refresh failed for 'antigravity_oauth_1.json' (timeout). Retry 2/3, back of queue.
+2026-01-24 17:06:59,936 - rotator_library - WARNING - Network error during refresh: [Errno -3] Temporary failure in name resolution, retry 1/3 in 1s
+2026-01-24 17:07:04,912 - rotator_library - WARNING - Refresh timeout (15s) for 'antigravity_oauth_1.json'
+2026-01-24 17:07:04,912 - rotator_library - ERROR - Max retries (3) reached for 'antigravity_oauth_1.json' (last error: timeout). Will retry next refresh cycle.

logs/proxy_debug.log

@@ -0,0 +1,679 @@
+2026-01-23 15:44:10,821 - rotator_library - DEBUG - ProviderCache[gemini3_signatures]: Disk enabled (memory_ttl=3600s, disk_ttl=86400s)
+2026-01-23 15:44:10,821 - rotator_library - DEBUG - ProviderCache[claude_thinking]: Disk enabled (memory_ttl=3600s, disk_ttl=86400s)
+2026-01-23 15:44:10,821 - rotator_library - DEBUG - Antigravity config: signatures_in_client=True, cache=True, dynamic_models=False, gemini3_fix=True, gemini3_strict_schema=True, claude_fix=False, thinking_sanitization=True, parallel_tool_claude=True, parallel_tool_gemini3=True
+2026-01-23 15:44:10,822 - rotator_library - DEBUG - Loaded persisted tier 'standard-tier' for credential: antigravity_oauth_1.json
+2026-01-23 15:44:10,822 - rotator_library - DEBUG - Loaded persisted tier 'standard-tier' for credential: antigravity_oauth_2.json
+2026-01-23 15:44:10,822 - rotator_library - DEBUG - antigravity: Loaded 2 credential tiers from disk: standard-tier=2
+2026-01-23 15:44:10,823 - rotator_library - DEBUG - Loading ANTIGRAVITY credentials from file: /app/oauth_creds/antigravity_oauth_1.json
+2026-01-23 15:44:10,823 - rotator_library - DEBUG - Loading ANTIGRAVITY credentials from file: /app/oauth_creds/antigravity_oauth_2.json
+2026-01-23 15:44:10,825 - rotator_library - DEBUG - ProviderCache[gemini3_signatures]: Started background tasks
+2026-01-23 15:44:10,825 - rotator_library - DEBUG - ProviderCache[claude_thinking]: Started background tasks
+2026-01-23 15:44:10,826 - rotator_library - DEBUG - Fetching quota baselines for 2 credentials...
+2026-01-23 15:44:13,440 - rotator_library - DEBUG - Baseline fetch complete: 2/2 successful
+2026-01-23 15:44:13,441 - rotator_library - DEBUG - ProviderCache[gemini3_signatures]: Disk enabled (memory_ttl=3600s, disk_ttl=86400s)
+2026-01-23 15:44:13,441 - rotator_library - DEBUG - ProviderCache[claude_thinking]: Disk enabled (memory_ttl=3600s, disk_ttl=86400s)
+2026-01-23 15:44:13,442 - rotator_library - DEBUG - Antigravity config: signatures_in_client=True, cache=True, dynamic_models=False, gemini3_fix=True, gemini3_strict_schema=True, claude_fix=False, thinking_sanitization=True, parallel_tool_claude=True, parallel_tool_gemini3=True
+2026-01-23 15:44:13,442 - rotator_library - DEBUG - Updated quota baseline for antigravity_oauth_1.json model=antigravity/gemini-2.5-flash: remaining=100.00%, synced_request_count=0
+2026-01-23 15:44:13,443 - rotator_library - DEBUG - Updated quota baseline for antigravity_oauth_1.json model=antigravity/claude-opus-4.5: remaining=100.00%, synced_request_count=0
+2026-01-23 15:44:13,444 - rotator_library - DEBUG - Updated quota baseline for antigravity_oauth_1.json model=antigravity/gemini-2.5-flash-lite: remaining=100.00%, synced_request_count=0
+2026-01-23 15:44:13,491 - rotator_library - DEBUG - Updated quota baseline for antigravity_oauth_1.json model=antigravity/gemini-3-pro-preview: remaining=100.00%, synced_request_count=0
+2026-01-23 15:44:13,492 - rotator_library - DEBUG - Updated quota baseline for antigravity_oauth_1.json model=antigravity/claude-sonnet-4.5: remaining=100.00%, synced_request_count=0
+2026-01-23 15:44:13,681 - rotator_library - DEBUG - Updated quota baseline for antigravity_oauth_1.json model=antigravity/gemini-3-flash: remaining=100.00%, synced_request_count=0
+2026-01-23 15:44:13,682 - rotator_library - DEBUG - Updated quota baseline for antigravity_oauth_2.json model=antigravity/gemini-2.5-flash-lite: remaining=100.00%, synced_request_count=0
+2026-01-23 15:44:13,861 - rotator_library - DEBUG - Updated quota baseline for antigravity_oauth_2.json model=antigravity/claude-sonnet-4.5: remaining=100.00%, synced_request_count=0
+2026-01-23 15:44:13,862 - rotator_library - DEBUG - Updated quota baseline for antigravity_oauth_2.json model=antigravity/gemini-2.5-flash: remaining=80.00%, synced_request_count=599
+2026-01-23 15:44:14,020 - rotator_library - DEBUG - Updated quota baseline for antigravity_oauth_2.json model=antigravity/gemini-3-pro-preview: remaining=100.00%, synced_request_count=0
+2026-01-23 15:44:14,021 - rotator_library - DEBUG - Updated quota baseline for antigravity_oauth_2.json model=antigravity/gemini-3-flash: remaining=80.00%, synced_request_count=79
+2026-01-23 15:44:14,168 - rotator_library - DEBUG - Updated quota baseline for antigravity_oauth_2.json model=antigravity/claude-opus-4.5: remaining=100.00%, synced_request_count=0
+2026-01-23 15:44:14,170 - rotator_library - DEBUG - Antigravity quota baseline refresh: no cooldowns needed
+2026-01-23 15:44:14,170 - rotator_library - DEBUG - antigravity quota refresh: updated 12 model baselines
+2026-01-23 15:44:14,170 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: initial run complete
+2026-01-23 15:44:14,170 - rotator_library - DEBUG - ProviderCache[gemini3_signatures]: Started background tasks
+2026-01-23 15:44:14,170 - rotator_library - DEBUG - ProviderCache[claude_thinking]: Started background tasks
+2026-01-23 15:44:28,311 - rotator_library - DEBUG - Handling reasoning parameters: model=antigravity/gemini-2.0-flash-exp, reasoning_effort=None
+2026-01-23 15:44:28,312 - rotator_library - DEBUG - Credential priorities for antigravity: P2=2
+2026-01-23 15:44:28,312 - rotator_library - DEBUG - Lazy-loaded tier 'standard-tier' for credential: antigravity_oauth_1.json
+2026-01-23 15:44:28,313 - rotator_library - DEBUG - Lazy-loaded tier 'standard-tier' for credential: antigravity_oauth_2.json
+2026-01-23 15:44:28,316 - rotator_library - DEBUG - Sequential ordering: antigravity_oauth_1.json(p=2, u=0) → antigravity_oauth_2.json(p=2, u=0)
+2026-01-23 15:44:28,317 - rotator_library - DEBUG - Provider 'antigravity' has custom logic. Delegating call.
+2026-01-23 15:44:28,318 - rotator_library - DEBUG - Starting Antigravity project discovery for credential: /app/oauth_creds/antigravity_oauth_1.json
+2026-01-23 15:44:28,318 - rotator_library - DEBUG - Using cached project ID: hopeful-inkwell-mwz46
+2026-01-23 15:44:48,448 - rotator_library - DEBUG - Provider 'antigravity' has custom logic. Delegating call.
+2026-01-23 15:44:48,449 - rotator_library - DEBUG - Starting Antigravity project discovery for credential: /app/oauth_creds/antigravity_oauth_2.json
+2026-01-23 15:44:48,449 - rotator_library - DEBUG - Using cached project ID: bold-throne-g2hmx
+2026-01-23 15:45:21,582 - rotator_library - DEBUG - Handling reasoning parameters: model=antigravity/gemini-1.5-pro, reasoning_effort=None
+2026-01-23 15:45:21,583 - rotator_library - DEBUG - Credential priorities for antigravity: P2=2
+2026-01-23 15:45:21,586 - rotator_library - DEBUG - Sequential ordering: antigravity_oauth_1.json(p=2, u=0) → antigravity_oauth_2.json(p=2, u=0)
+2026-01-23 15:45:21,586 - rotator_library - DEBUG - Provider 'antigravity' has custom logic. Delegating call.
+2026-01-23 15:45:21,587 - rotator_library - DEBUG - Starting Antigravity project discovery for credential: /app/oauth_creds/antigravity_oauth_1.json
+2026-01-23 15:45:21,587 - rotator_library - DEBUG - Using cached project ID: hopeful-inkwell-mwz46
+2026-01-23 15:45:37,865 - rotator_library - DEBUG - Provider 'antigravity' has custom logic. Delegating call.
+2026-01-23 15:45:37,866 - rotator_library - DEBUG - Starting Antigravity project discovery for credential: /app/oauth_creds/antigravity_oauth_2.json
+2026-01-23 15:45:37,866 - rotator_library - DEBUG - Using cached project ID: bold-throne-g2hmx
+2026-01-23 15:49:14,171 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-23 15:49:14,171 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-23 15:54:14,172 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-23 15:54:14,172 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-23 15:56:23,673 - rotator_library - DEBUG - Handling reasoning parameters: model=antigravity/gemini-2.0-flash-exp, reasoning_effort=None
+2026-01-23 15:56:23,674 - rotator_library - DEBUG - Credential priorities for antigravity: P2=2
+2026-01-23 15:56:23,676 - rotator_library - DEBUG - Sequential ordering: antigravity_oauth_1.json(p=2, u=1) → antigravity_oauth_2.json(p=2, u=1)
+2026-01-23 15:56:23,677 - rotator_library - DEBUG - Provider 'antigravity' has custom logic. Delegating call.
+2026-01-23 15:56:23,678 - rotator_library - DEBUG - Starting Antigravity project discovery for credential: /app/oauth_creds/antigravity_oauth_1.json
+2026-01-23 15:56:23,678 - rotator_library - DEBUG - Using cached project ID: hopeful-inkwell-mwz46
+2026-01-23 15:56:41,164 - rotator_library - DEBUG - Provider 'antigravity' has custom logic. Delegating call.
+2026-01-23 15:56:41,164 - rotator_library - DEBUG - Starting Antigravity project discovery for credential: /app/oauth_creds/antigravity_oauth_2.json
+2026-01-23 15:56:41,164 - rotator_library - DEBUG - Using cached project ID: bold-throne-g2hmx
+2026-01-23 15:58:31,359 - rotator_library - DEBUG - Attempting to get models for antigravity with credential antigravity_oauth_2.json
+2026-01-23 15:58:31,359 - rotator_library - DEBUG - Using hardcoded model list
+2026-01-23 15:59:07,975 - rotator_library - DEBUG - Handling reasoning parameters: model=antigravity/gemini-3-flash, reasoning_effort=None
+2026-01-23 15:59:07,976 - rotator_library - DEBUG - Credential priorities for antigravity: P2=2
+2026-01-23 15:59:07,978 - rotator_library - DEBUG - Sequential ordering: antigravity_oauth_2.json(p=2, u=79) → antigravity_oauth_1.json(p=2, u=0)
+2026-01-23 15:59:07,978 - rotator_library - DEBUG - Provider 'antigravity' has custom logic. Delegating call.
+2026-01-23 15:59:07,979 - rotator_library - DEBUG - Starting Antigravity project discovery for credential: /app/oauth_creds/antigravity_oauth_2.json
+2026-01-23 15:59:07,979 - rotator_library - DEBUG - Using cached project ID: bold-throne-g2hmx
+2026-01-23 15:59:14,173 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-23 15:59:14,173 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-23 15:59:24,453 - rotator_library - DEBUG - Provider 'antigravity' has custom logic. Delegating call.
+2026-01-23 15:59:24,453 - rotator_library - DEBUG - Starting Antigravity project discovery for credential: /app/oauth_creds/antigravity_oauth_1.json
+2026-01-23 15:59:24,454 - rotator_library - DEBUG - Using cached project ID: hopeful-inkwell-mwz46
+2026-01-23 16:00:10,954 - rotator_library - DEBUG - Returning cached models for provider: antigravity
+2026-01-23 16:00:21,201 - rotator_library - DEBUG - Returning cached models for provider: antigravity
+2026-01-23 16:01:08,173 - rotator_library - DEBUG - Handling reasoning parameters: model=antigravity/gemini-2.5-flash, reasoning_effort=None
+2026-01-23 16:01:08,173 - rotator_library - DEBUG - Credential priorities for antigravity: P2=2
+2026-01-23 16:01:08,175 - rotator_library - DEBUG - Sequential ordering: antigravity_oauth_2.json(p=2, u=599) → antigravity_oauth_1.json(p=2, u=0)
+2026-01-23 16:01:08,176 - rotator_library - DEBUG - Provider 'antigravity' has custom logic. Delegating call.
+2026-01-23 16:01:08,176 - rotator_library - DEBUG - Starting Antigravity project discovery for credential: /app/oauth_creds/antigravity_oauth_2.json
+2026-01-23 16:01:08,176 - rotator_library - DEBUG - Using cached project ID: bold-throne-g2hmx
+2026-01-23 16:01:26,091 - rotator_library - DEBUG - Provider 'antigravity' has custom logic. Delegating call.
+2026-01-23 16:01:26,092 - rotator_library - DEBUG - Starting Antigravity project discovery for credential: /app/oauth_creds/antigravity_oauth_1.json
+2026-01-23 16:01:26,092 - rotator_library - DEBUG - Using cached project ID: hopeful-inkwell-mwz46
+2026-01-23 16:02:46,115 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: cancelled
+2026-01-23 16:02:46,115 - rotator_library - DEBUG - Stopped background job for 'antigravity'
+2026-01-23 16:03:05,746 - rotator_library - DEBUG - ProviderCache[gemini3_signatures]: Disk enabled (memory_ttl=3600s, disk_ttl=86400s)
+2026-01-23 16:03:05,747 - rotator_library - DEBUG - ProviderCache[claude_thinking]: Disk enabled (memory_ttl=3600s, disk_ttl=86400s)
+2026-01-23 16:03:05,747 - rotator_library - DEBUG - Antigravity config: signatures_in_client=True, cache=True, dynamic_models=False, gemini3_fix=True, gemini3_strict_schema=True, claude_fix=False, thinking_sanitization=True, parallel_tool_claude=True, parallel_tool_gemini3=True
+2026-01-23 16:03:05,747 - rotator_library - DEBUG - Loaded persisted tier 'standard-tier' for credential: antigravity_oauth_1.json
+2026-01-23 16:03:05,748 - rotator_library - DEBUG - Loaded persisted tier 'standard-tier' for credential: antigravity_oauth_2.json
+2026-01-23 16:03:05,748 - rotator_library - DEBUG - antigravity: Loaded 2 credential tiers from disk: standard-tier=2
+2026-01-23 16:03:05,749 - rotator_library - DEBUG - Loading ANTIGRAVITY credentials from file: /app/oauth_creds/antigravity_oauth_1.json
+2026-01-23 16:03:05,749 - rotator_library - DEBUG - Loading ANTIGRAVITY credentials from file: /app/oauth_creds/antigravity_oauth_2.json
+2026-01-23 16:03:05,751 - rotator_library - DEBUG - ProviderCache[gemini3_signatures]: Started background tasks
+2026-01-23 16:03:05,751 - rotator_library - DEBUG - ProviderCache[claude_thinking]: Started background tasks
+2026-01-23 16:03:05,751 - rotator_library - DEBUG - Fetching quota baselines for 2 credentials...
+2026-01-23 16:03:08,351 - rotator_library - DEBUG - Baseline fetch complete: 2/2 successful
+2026-01-23 16:03:08,353 - rotator_library - DEBUG - ProviderCache[gemini3_signatures]: Disk enabled (memory_ttl=3600s, disk_ttl=86400s)
+2026-01-23 16:03:08,354 - rotator_library - DEBUG - ProviderCache[claude_thinking]: Disk enabled (memory_ttl=3600s, disk_ttl=86400s)
+2026-01-23 16:03:08,354 - rotator_library - DEBUG - Antigravity config: signatures_in_client=True, cache=True, dynamic_models=False, gemini3_fix=True, gemini3_strict_schema=True, claude_fix=False, thinking_sanitization=True, parallel_tool_claude=True, parallel_tool_gemini3=True
+2026-01-23 16:03:08,355 - rotator_library - DEBUG - Lazy-loaded tier 'standard-tier' for credential: antigravity_oauth_1.json
+2026-01-23 16:03:08,356 - rotator_library - DEBUG - Lazy-loaded tier 'standard-tier' for credential: antigravity_oauth_2.json
+2026-01-23 16:03:08,356 - rotator_library - DEBUG - Updated quota baseline for antigravity_oauth_1.json model=antigravity/claude-sonnet-4.5: remaining=100.00%, synced_request_count=0
+2026-01-23 16:03:08,358 - rotator_library - DEBUG - Updated quota baseline for antigravity_oauth_1.json model=antigravity/gemini-2.5-flash-lite: remaining=100.00%, synced_request_count=0
+2026-01-23 16:03:08,360 - rotator_library - DEBUG - Updated quota baseline for antigravity_oauth_1.json model=antigravity/gemini-3-flash: remaining=100.00%, synced_request_count=1
+2026-01-23 16:03:08,360 - rotator_library - DEBUG - Updated quota baseline for antigravity_oauth_1.json model=antigravity/gemini-3-pro-preview: remaining=100.00%, synced_request_count=0
+2026-01-23 16:03:08,361 - rotator_library - DEBUG - Updated quota baseline for antigravity_oauth_1.json model=antigravity/gemini-2.5-flash: remaining=100.00%, synced_request_count=1
+2026-01-23 16:03:08,363 - rotator_library - DEBUG - Updated quota baseline for antigravity_oauth_1.json model=antigravity/claude-opus-4.5: remaining=100.00%, synced_request_count=0
+2026-01-23 16:03:08,364 - rotator_library - DEBUG - Updated quota baseline for antigravity_oauth_2.json model=antigravity/claude-sonnet-4.5: remaining=100.00%, synced_request_count=0
+2026-01-23 16:03:08,365 - rotator_library - DEBUG - Updated quota baseline for antigravity_oauth_2.json model=antigravity/gemini-2.5-flash: remaining=80.00%, synced_request_count=600
+2026-01-23 16:03:08,366 - rotator_library - DEBUG - Updated quota baseline for antigravity_oauth_2.json model=antigravity/gemini-2.5-flash-lite: remaining=100.00%, synced_request_count=0
+2026-01-23 16:03:08,367 - rotator_library - DEBUG - Updated quota baseline for antigravity_oauth_2.json model=antigravity/gemini-3-pro-preview: remaining=100.00%, synced_request_count=0
+2026-01-23 16:03:08,368 - rotator_library - DEBUG - Updated quota baseline for antigravity_oauth_2.json model=antigravity/gemini-3-flash: remaining=80.00%, synced_request_count=80
+2026-01-23 16:03:08,369 - rotator_library - DEBUG - Updated quota baseline for antigravity_oauth_2.json model=antigravity/claude-opus-4.5: remaining=100.00%, synced_request_count=0
+2026-01-23 16:03:08,370 - rotator_library - DEBUG - Antigravity quota baseline refresh: no cooldowns needed
+2026-01-23 16:03:08,370 - rotator_library - DEBUG - antigravity quota refresh: updated 12 model baselines
+2026-01-23 16:03:08,370 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: initial run complete
+2026-01-23 16:03:08,370 - rotator_library - DEBUG - ProviderCache[gemini3_signatures]: Started background tasks
+2026-01-23 16:03:08,370 - rotator_library - DEBUG - ProviderCache[claude_thinking]: Started background tasks
+2026-01-23 16:04:06,561 - rotator_library - DEBUG - Attempting to get models for antigravity with credential antigravity_oauth_1.json
+2026-01-23 16:04:06,564 - rotator_library - DEBUG - Using hardcoded model list
+2026-01-23 16:04:22,953 - rotator_library - DEBUG - Handling reasoning parameters: model=antigravity/gemini-3-flash, reasoning_effort=None
+2026-01-23 16:04:22,954 - rotator_library - DEBUG - Credential priorities for antigravity: P2=2
+2026-01-23 16:04:22,957 - rotator_library - DEBUG - Sequential ordering: antigravity_oauth_2.json(p=2, u=80) → antigravity_oauth_1.json(p=2, u=1)
+2026-01-23 16:04:22,958 - rotator_library - DEBUG - Provider 'antigravity' has custom logic. Delegating call.
+2026-01-23 16:04:22,958 - rotator_library - DEBUG - Starting Antigravity project discovery for credential: /app/oauth_creds/antigravity_oauth_2.json
+2026-01-23 16:04:22,959 - rotator_library - DEBUG - Using cached project ID: bold-throne-g2hmx
+2026-01-23 16:04:25,936 - rotator_library - DEBUG - Skipping cost calculation for provider 'antigravity' (custom provider).
+2026-01-23 16:08:08,371 - rotator_library - DEBUG - Refreshing quota baselines for 1 recently active credentials
+2026-01-23 16:08:10,408 - rotator_library - DEBUG - Updated quota baseline for antigravity_oauth_2.json model=antigravity/gemini-3-flash: remaining=80.00%, synced_request_count=81
+2026-01-23 16:08:10,409 - rotator_library - DEBUG - Updated quota baseline for antigravity_oauth_2.json model=antigravity/gemini-2.5-flash: remaining=80.00%, synced_request_count=600
+2026-01-23 16:08:10,410 - rotator_library - DEBUG - Updated quota baseline for antigravity_oauth_2.json model=antigravity/claude-opus-4.5: remaining=100.00%, synced_request_count=0
+2026-01-23 16:08:10,412 - rotator_library - DEBUG - Updated quota baseline for antigravity_oauth_2.json model=antigravity/gemini-3-pro-preview: remaining=100.00%, synced_request_count=0
+2026-01-23 16:08:10,416 - rotator_library - DEBUG - Updated quota baseline for antigravity_oauth_2.json model=antigravity/claude-sonnet-4.5: remaining=100.00%, synced_request_count=0
+2026-01-23 16:08:10,417 - rotator_library - DEBUG - Updated quota baseline for antigravity_oauth_2.json model=antigravity/gemini-2.5-flash-lite: remaining=100.00%, synced_request_count=0
+2026-01-23 16:08:10,418 - rotator_library - DEBUG - Antigravity quota baseline refresh: no cooldowns needed
+2026-01-23 16:08:10,418 - rotator_library - DEBUG - antigravity quota refresh: updated 6 model baselines
+2026-01-23 16:08:10,418 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-23 16:12:00,819 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: cancelled
+2026-01-23 16:12:00,819 - rotator_library - DEBUG - Stopped background job for 'antigravity'
+2026-01-23 16:12:06,403 - rotator_library - DEBUG - ProviderCache[gemini3_signatures]: Disk enabled (memory_ttl=3600s, disk_ttl=86400s)
+2026-01-23 16:12:06,404 - rotator_library - DEBUG - ProviderCache[claude_thinking]: Disk enabled (memory_ttl=3600s, disk_ttl=86400s)
+2026-01-23 16:12:06,404 - rotator_library - DEBUG - Antigravity config: signatures_in_client=True, cache=True, dynamic_models=False, gemini3_fix=True, gemini3_strict_schema=True, claude_fix=False, thinking_sanitization=True, parallel_tool_claude=True, parallel_tool_gemini3=True
+2026-01-23 16:12:06,404 - rotator_library - DEBUG - Loaded persisted tier 'standard-tier' for credential: antigravity_oauth_1.json
+2026-01-23 16:12:06,404 - rotator_library - DEBUG - antigravity: Loaded 1 credential tiers from disk: standard-tier=1
+2026-01-23 16:12:06,405 - rotator_library - DEBUG - Loading ANTIGRAVITY credentials from file: /app/oauth_creds/antigravity_oauth_1.json
+2026-01-23 16:12:06,407 - rotator_library - DEBUG - ProviderCache[gemini3_signatures]: Started background tasks
+2026-01-23 16:12:06,407 - rotator_library - DEBUG - ProviderCache[claude_thinking]: Started background tasks
+2026-01-23 16:12:06,407 - rotator_library - DEBUG - Fetching quota baselines for 1 credentials...
+2026-01-23 16:12:09,041 - rotator_library - DEBUG - Baseline fetch complete: 1/1 successful
+2026-01-23 16:12:09,043 - rotator_library - DEBUG - ProviderCache[gemini3_signatures]: Disk enabled (memory_ttl=3600s, disk_ttl=86400s)
+2026-01-23 16:12:09,043 - rotator_library - DEBUG - ProviderCache[claude_thinking]: Disk enabled (memory_ttl=3600s, disk_ttl=86400s)
+2026-01-23 16:12:09,043 - rotator_library - DEBUG - Antigravity config: signatures_in_client=True, cache=True, dynamic_models=False, gemini3_fix=True, gemini3_strict_schema=True, claude_fix=False, thinking_sanitization=True, parallel_tool_claude=True, parallel_tool_gemini3=True
+2026-01-23 16:12:09,044 - rotator_library - DEBUG - Lazy-loaded tier 'standard-tier' for credential: antigravity_oauth_1.json
+2026-01-23 16:12:09,045 - rotator_library - DEBUG - Could not lazy-load tier from /app/oauth_creds/antigravity_oauth_2.json: [Errno 2] No such file or directory: '/app/oauth_creds/antigravity_oauth_2.json'
+2026-01-23 16:12:09,045 - rotator_library - DEBUG - Updated quota baseline for antigravity_oauth_1.json model=antigravity/gemini-2.5-flash: remaining=100.00%, synced_request_count=1
+2026-01-23 16:12:09,051 - rotator_library - DEBUG - Updated quota baseline for antigravity_oauth_1.json model=antigravity/gemini-3-flash: remaining=100.00%, synced_request_count=1
+2026-01-23 16:12:09,678 - rotator_library - DEBUG - Updated quota baseline for antigravity_oauth_1.json model=antigravity/claude-sonnet-4.5: remaining=100.00%, synced_request_count=0
+2026-01-23 16:12:09,679 - rotator_library - DEBUG - Updated quota baseline for antigravity_oauth_1.json model=antigravity/gemini-3-pro-preview: remaining=100.00%, synced_request_count=0
+2026-01-23 16:12:09,837 - rotator_library - DEBUG - Updated quota baseline for antigravity_oauth_1.json model=antigravity/claude-opus-4.5: remaining=100.00%, synced_request_count=0
+2026-01-23 16:12:09,839 - rotator_library - DEBUG - Updated quota baseline for antigravity_oauth_1.json model=antigravity/gemini-2.5-flash-lite: remaining=100.00%, synced_request_count=0
+2026-01-23 16:12:09,973 - rotator_library - DEBUG - Antigravity quota baseline refresh: no cooldowns needed
+2026-01-23 16:12:09,973 - rotator_library - DEBUG - antigravity quota refresh: updated 6 model baselines
+2026-01-23 16:12:09,973 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: initial run complete
+2026-01-23 16:12:09,974 - rotator_library - DEBUG - ProviderCache[gemini3_signatures]: Started background tasks
+2026-01-23 16:12:09,974 - rotator_library - DEBUG - ProviderCache[claude_thinking]: Started background tasks
+2026-01-23 16:17:09,974 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-23 16:17:09,974 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-23 16:22:06,407 - rotator_library - DEBUG - Refreshing ANTIGRAVITY OAuth token for 'antigravity_oauth_1.json' (forced: False)...
+2026-01-23 16:22:07,224 - rotator_library - DEBUG - Token validation successful for 'antigravity_oauth_1.json'
+2026-01-23 16:22:07,226 - rotator_library - DEBUG - Saved updated ANTIGRAVITY OAuth credentials to '/app/oauth_creds/antigravity_oauth_1.json'.
+2026-01-23 16:22:07,226 - rotator_library - DEBUG - Successfully refreshed ANTIGRAVITY OAuth token for 'antigravity_oauth_1.json'.
+2026-01-23 16:22:09,975 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-23 16:22:09,976 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-23 16:27:09,977 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-23 16:27:09,977 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-23 16:32:09,978 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-23 16:32:09,979 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-23 16:37:09,980 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-23 16:37:09,980 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-23 16:38:58,754 - rotator_library - DEBUG - Handling reasoning parameters: model=antigravity/gemini-3-flash, reasoning_effort=None
+2026-01-23 16:38:58,755 - rotator_library - DEBUG - Credential priorities for antigravity: P2=1
+2026-01-23 16:38:58,756 - rotator_library - DEBUG - Could not lazy-load tier from /app/oauth_creds/antigravity_oauth_2.json: [Errno 2] No such file or directory: '/app/oauth_creds/antigravity_oauth_2.json'
+2026-01-23 16:38:58,757 - rotator_library - DEBUG - Provider 'antigravity' has custom logic. Delegating call.
+2026-01-23 16:38:58,758 - rotator_library - DEBUG - Starting Antigravity project discovery for credential: /app/oauth_creds/antigravity_oauth_1.json
+2026-01-23 16:38:58,758 - rotator_library - DEBUG - Using cached project ID: hopeful-inkwell-mwz46
+2026-01-23 16:39:02,594 - rotator_library - DEBUG - Skipping cost calculation for provider 'antigravity' (custom provider).
+2026-01-23 16:42:09,981 - rotator_library - DEBUG - Refreshing quota baselines for 1 recently active credentials
+2026-01-23 16:42:12,436 - rotator_library - DEBUG - Updated quota baseline for antigravity_oauth_1.json model=antigravity/claude-sonnet-4.5: remaining=100.00%, synced_request_count=0
+2026-01-23 16:42:12,438 - rotator_library - DEBUG - Updated quota baseline for antigravity_oauth_1.json model=antigravity/gemini-3-flash: remaining=100.00%, synced_request_count=2
+2026-01-23 16:42:12,440 - rotator_library - DEBUG - Updated quota baseline for antigravity_oauth_1.json model=antigravity/gemini-2.5-flash-lite: remaining=100.00%, synced_request_count=0
+2026-01-23 16:42:12,444 - rotator_library - DEBUG - Updated quota baseline for antigravity_oauth_1.json model=antigravity/gemini-2.5-flash: remaining=100.00%, synced_request_count=1
+2026-01-23 16:42:12,446 - rotator_library - DEBUG - Updated quota baseline for antigravity_oauth_1.json model=antigravity/claude-opus-4.5: remaining=100.00%, synced_request_count=0
+2026-01-23 16:42:12,448 - rotator_library - DEBUG - Updated quota baseline for antigravity_oauth_1.json model=antigravity/gemini-3-pro-preview: remaining=100.00%, synced_request_count=0
+2026-01-23 16:42:12,450 - rotator_library - DEBUG - Antigravity quota baseline refresh: no cooldowns needed
+2026-01-23 16:42:12,450 - rotator_library - DEBUG - antigravity quota refresh: updated 6 model baselines
+2026-01-23 16:42:12,451 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-23 16:47:12,452 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-23 16:47:12,452 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-23 16:52:06,409 - rotator_library - DEBUG - Refreshing ANTIGRAVITY OAuth token for 'antigravity_oauth_1.json' (forced: False)...
+2026-01-23 16:52:07,231 - rotator_library - DEBUG - Token validation successful for 'antigravity_oauth_1.json'
+2026-01-23 16:52:07,233 - rotator_library - DEBUG - Saved updated ANTIGRAVITY OAuth credentials to '/app/oauth_creds/antigravity_oauth_1.json'.
+2026-01-23 16:52:07,233 - rotator_library - DEBUG - Successfully refreshed ANTIGRAVITY OAuth token for 'antigravity_oauth_1.json'.
+2026-01-23 16:52:12,453 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-23 16:52:12,454 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 03:36:39,099 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 03:36:39,099 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 03:41:33,054 - rotator_library - DEBUG - Refreshing ANTIGRAVITY OAuth token for 'antigravity_oauth_1.json' (forced: False)...
+2026-01-24 03:41:33,814 - rotator_library - DEBUG - Token validation successful for 'antigravity_oauth_1.json'
+2026-01-24 03:41:33,816 - rotator_library - DEBUG - Saved updated ANTIGRAVITY OAuth credentials to '/app/oauth_creds/antigravity_oauth_1.json'.
+2026-01-24 03:41:33,816 - rotator_library - DEBUG - Successfully refreshed ANTIGRAVITY OAuth token for 'antigravity_oauth_1.json'.
+2026-01-24 03:41:39,102 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 03:41:39,102 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 03:46:39,103 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 03:46:39,103 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 03:51:39,104 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 03:51:39,104 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 03:56:39,105 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 03:56:39,105 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 04:01:39,107 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 04:01:39,107 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 04:06:39,108 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 04:06:39,108 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 07:04:03,801 - rotator_library - DEBUG - Refreshing ANTIGRAVITY OAuth token for 'antigravity_oauth_1.json' (forced: False)...
+2026-01-24 07:04:04,545 - rotator_library - DEBUG - Token validation successful for 'antigravity_oauth_1.json'
+2026-01-24 07:04:04,547 - rotator_library - DEBUG - Saved updated ANTIGRAVITY OAuth credentials to '/app/oauth_creds/antigravity_oauth_1.json'.
+2026-01-24 07:04:04,547 - rotator_library - DEBUG - Successfully refreshed ANTIGRAVITY OAuth token for 'antigravity_oauth_1.json'.
+2026-01-24 07:04:09,855 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 07:04:09,855 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 07:09:09,856 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 07:09:09,856 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 07:14:09,857 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 07:14:09,857 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 07:19:15,130 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 07:19:15,130 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 07:24:15,131 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 07:24:15,131 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 07:29:15,132 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 07:29:15,132 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 07:35:14,550 - rotator_library - DEBUG - Refreshing ANTIGRAVITY OAuth token for 'antigravity_oauth_1.json' (forced: False)...
+2026-01-24 07:35:15,430 - rotator_library - DEBUG - Token validation successful for 'antigravity_oauth_1.json'
+2026-01-24 07:35:15,432 - rotator_library - DEBUG - Saved updated ANTIGRAVITY OAuth credentials to '/app/oauth_creds/antigravity_oauth_1.json'.
+2026-01-24 07:35:15,432 - rotator_library - DEBUG - Successfully refreshed ANTIGRAVITY OAuth token for 'antigravity_oauth_1.json'.
+2026-01-24 07:35:20,607 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 07:35:20,608 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 07:40:20,610 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 07:40:20,610 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 07:45:20,612 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 07:45:20,612 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 07:50:20,613 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 07:50:20,613 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 07:55:20,615 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 07:55:20,615 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 08:00:20,616 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 08:00:20,616 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 08:05:14,554 - rotator_library - DEBUG - Refreshing ANTIGRAVITY OAuth token for 'antigravity_oauth_1.json' (forced: False)...
+2026-01-24 08:05:15,298 - rotator_library - DEBUG - Token validation successful for 'antigravity_oauth_1.json'
+2026-01-24 08:05:15,300 - rotator_library - DEBUG - Saved updated ANTIGRAVITY OAuth credentials to '/app/oauth_creds/antigravity_oauth_1.json'.
+2026-01-24 08:05:15,300 - rotator_library - DEBUG - Successfully refreshed ANTIGRAVITY OAuth token for 'antigravity_oauth_1.json'.
+2026-01-24 08:05:20,617 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 08:05:20,618 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 08:10:20,619 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 08:10:20,619 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 08:15:20,620 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 08:15:20,621 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 08:20:20,621 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 08:20:20,622 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 08:25:20,622 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 08:25:20,623 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 08:30:20,623 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 08:30:20,624 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 08:35:14,557 - rotator_library - DEBUG - Refreshing ANTIGRAVITY OAuth token for 'antigravity_oauth_1.json' (forced: False)...
+2026-01-24 08:35:15,319 - rotator_library - DEBUG - Token validation successful for 'antigravity_oauth_1.json'
+2026-01-24 08:35:15,541 - rotator_library - DEBUG - Saved updated ANTIGRAVITY OAuth credentials to '/app/oauth_creds/antigravity_oauth_1.json'.
+2026-01-24 08:35:15,541 - rotator_library - DEBUG - Successfully refreshed ANTIGRAVITY OAuth token for 'antigravity_oauth_1.json'.
+2026-01-24 08:35:20,624 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 08:35:20,624 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 08:40:20,625 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 08:40:20,625 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 08:45:20,626 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 08:45:20,627 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 08:50:20,627 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 08:50:21,005 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 08:55:21,005 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 08:55:21,006 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 09:00:21,006 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 09:00:21,008 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 09:05:14,560 - rotator_library - DEBUG - Refreshing ANTIGRAVITY OAuth token for 'antigravity_oauth_1.json' (forced: False)...
+2026-01-24 09:05:15,874 - rotator_library - DEBUG - Token validation successful for 'antigravity_oauth_1.json'
+2026-01-24 09:05:16,123 - rotator_library - DEBUG - Saved updated ANTIGRAVITY OAuth credentials to '/app/oauth_creds/antigravity_oauth_1.json'.
+2026-01-24 09:05:16,123 - rotator_library - DEBUG - Successfully refreshed ANTIGRAVITY OAuth token for 'antigravity_oauth_1.json'.
+2026-01-24 09:05:21,009 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 09:05:21,009 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 09:10:21,010 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 09:10:21,010 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 09:15:21,011 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 09:15:21,011 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 09:20:21,012 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 09:20:21,012 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 09:25:21,013 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 09:25:21,014 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 09:30:21,015 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 09:30:21,015 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 09:35:14,562 - rotator_library - DEBUG - Refreshing ANTIGRAVITY OAuth token for 'antigravity_oauth_1.json' (forced: False)...
+2026-01-24 09:35:15,530 - rotator_library - DEBUG - Token validation successful for 'antigravity_oauth_1.json'
+2026-01-24 09:35:15,619 - rotator_library - DEBUG - Saved updated ANTIGRAVITY OAuth credentials to '/app/oauth_creds/antigravity_oauth_1.json'.
+2026-01-24 09:35:15,619 - rotator_library - DEBUG - Successfully refreshed ANTIGRAVITY OAuth token for 'antigravity_oauth_1.json'.
+2026-01-24 09:35:21,016 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 09:35:21,016 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 09:40:21,087 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 09:40:21,088 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 09:45:21,088 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 09:45:21,091 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 09:50:21,091 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 09:50:21,091 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 09:55:21,093 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 09:55:21,093 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 10:00:21,094 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 10:00:21,094 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 10:05:14,568 - rotator_library - DEBUG - Refreshing ANTIGRAVITY OAuth token for 'antigravity_oauth_1.json' (forced: False)...
+2026-01-24 10:05:15,410 - rotator_library - DEBUG - Token validation successful for 'antigravity_oauth_1.json'
+2026-01-24 10:05:15,460 - rotator_library - DEBUG - Saved updated ANTIGRAVITY OAuth credentials to '/app/oauth_creds/antigravity_oauth_1.json'.
+2026-01-24 10:05:15,460 - rotator_library - DEBUG - Successfully refreshed ANTIGRAVITY OAuth token for 'antigravity_oauth_1.json'.
+2026-01-24 10:05:21,095 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 10:05:21,096 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 10:10:21,097 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 10:10:21,097 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 10:15:21,101 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 10:15:21,101 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 10:20:21,102 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 10:20:21,102 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 10:25:21,103 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 10:25:21,103 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 10:30:21,104 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 10:30:21,105 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 10:35:14,571 - rotator_library - DEBUG - Refreshing ANTIGRAVITY OAuth token for 'antigravity_oauth_1.json' (forced: False)...
+2026-01-24 10:35:15,333 - rotator_library - DEBUG - Token validation successful for 'antigravity_oauth_1.json'
+2026-01-24 10:35:15,531 - rotator_library - DEBUG - Saved updated ANTIGRAVITY OAuth credentials to '/app/oauth_creds/antigravity_oauth_1.json'.
+2026-01-24 10:35:15,531 - rotator_library - DEBUG - Successfully refreshed ANTIGRAVITY OAuth token for 'antigravity_oauth_1.json'.
+2026-01-24 10:35:21,106 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 10:35:21,107 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 10:40:21,108 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 10:40:21,108 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 10:45:21,109 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 10:45:21,109 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 10:50:21,110 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 10:50:21,110 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 10:55:21,110 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 10:55:21,111 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 11:00:21,113 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 11:00:21,115 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 11:05:14,575 - rotator_library - DEBUG - Refreshing ANTIGRAVITY OAuth token for 'antigravity_oauth_1.json' (forced: False)...
+2026-01-24 11:05:15,395 - rotator_library - DEBUG - Token validation successful for 'antigravity_oauth_1.json'
+2026-01-24 11:05:15,397 - rotator_library - DEBUG - Saved updated ANTIGRAVITY OAuth credentials to '/app/oauth_creds/antigravity_oauth_1.json'.
+2026-01-24 11:05:15,397 - rotator_library - DEBUG - Successfully refreshed ANTIGRAVITY OAuth token for 'antigravity_oauth_1.json'.
+2026-01-24 11:05:21,116 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 11:05:21,116 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 11:10:21,117 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 11:10:21,118 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 11:15:21,119 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 11:15:21,120 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 11:18:44,059 - rotator_library - DEBUG - Attempting to get models for antigravity with credential antigravity_oauth_1.json
+2026-01-24 11:18:44,059 - rotator_library - DEBUG - Using hardcoded model list
+2026-01-24 11:20:21,121 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 11:20:21,121 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 11:23:14,659 - rotator_library - DEBUG - Returning cached models for provider: antigravity
+2026-01-24 11:25:21,121 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 11:25:21,122 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 11:30:21,122 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 11:30:21,123 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 11:32:50,245 - rotator_library - DEBUG - Handling reasoning parameters: model=antigravity/gemini-3-flash, reasoning_effort=None
+2026-01-24 11:32:50,411 - rotator_library - DEBUG - Credential priorities for antigravity: P2=1
+2026-01-24 11:32:50,524 - rotator_library - DEBUG - Could not lazy-load tier from /app/oauth_creds/antigravity_oauth_2.json: [Errno 2] No such file or directory: '/app/oauth_creds/antigravity_oauth_2.json'
+2026-01-24 11:32:50,881 - rotator_library - DEBUG - Provider 'antigravity' has custom logic. Delegating call.
+2026-01-24 11:32:50,882 - rotator_library - DEBUG - Starting Antigravity project discovery for credential: /app/oauth_creds/antigravity_oauth_1.json
+2026-01-24 11:32:50,883 - rotator_library - DEBUG - Using cached project ID: hopeful-inkwell-mwz46
+2026-01-24 11:32:51,589 - rotator_library - DEBUG - Handling reasoning parameters: model=antigravity/gemini-3-flash, reasoning_effort=None
+2026-01-24 11:32:51,591 - rotator_library - DEBUG - Credential priorities for antigravity: P2=1
+2026-01-24 11:32:51,598 - rotator_library - DEBUG - Could not lazy-load tier from /app/oauth_creds/antigravity_oauth_2.json: [Errno 2] No such file or directory: '/app/oauth_creds/antigravity_oauth_2.json'
+2026-01-24 11:32:51,599 - rotator_library - DEBUG - Provider 'antigravity' has custom logic. Delegating call.
+2026-01-24 11:32:51,599 - rotator_library - DEBUG - Starting Antigravity project discovery for credential: /app/oauth_creds/antigravity_oauth_1.json
+2026-01-24 11:32:51,599 - rotator_library - DEBUG - Using cached project ID: hopeful-inkwell-mwz46
+2026-01-24 11:32:51,607 - rotator_library - DEBUG - Handling reasoning parameters: model=antigravity/claude-opus-4.5, reasoning_effort=None
+2026-01-24 11:32:51,608 - rotator_library - DEBUG - Credential priorities for antigravity: P2=1
+2026-01-24 11:32:51,609 - rotator_library - DEBUG - Could not lazy-load tier from /app/oauth_creds/antigravity_oauth_2.json: [Errno 2] No such file or directory: '/app/oauth_creds/antigravity_oauth_2.json'
+2026-01-24 11:32:51,610 - rotator_library - DEBUG - Provider 'antigravity' has custom logic. Delegating call.
+2026-01-24 11:32:51,610 - rotator_library - DEBUG - [Thinking Sanitization] thinking_enabled=True, in_tool_loop=False, turn_has_thinking=False, turn_start_idx=-1
+2026-01-24 11:32:51,610 - rotator_library - DEBUG - [Interleaved Thinking] Injected reminder to user message at index 0
+2026-01-24 11:32:51,611 - rotator_library - DEBUG - [Schema] Preserving property 'pattern' (matches validation keyword name)
+2026-01-24 11:32:51,611 - rotator_library - DEBUG - [Schema] Preserving property 'pattern' (matches validation keyword name)
+2026-01-24 11:32:51,611 - rotator_library - DEBUG - [Schema] Preserving property 'format' (matches validation keyword name)
+2026-01-24 11:32:51,612 - rotator_library - DEBUG - [Schema] Preserving property 'pattern' (matches validation keyword name)
+2026-01-24 11:32:51,612 - rotator_library - DEBUG - [Schema] Preserving property 'pattern' (matches validation keyword name)
+2026-01-24 11:32:51,612 - rotator_library - DEBUG - Starting Antigravity project discovery for credential: /app/oauth_creds/antigravity_oauth_1.json
+2026-01-24 11:32:51,612 - rotator_library - DEBUG - Using cached project ID: hopeful-inkwell-mwz46
+2026-01-24 11:32:54,552 - rotator_library - DEBUG - Skipping cost calculation for provider 'antigravity' (custom provider).
+2026-01-24 11:32:56,228 - rotator_library - DEBUG - Skipping cost calculation for provider 'antigravity' (custom provider).
+2026-01-24 11:32:57,435 - rotator_library - DEBUG - Skipping cost calculation for provider 'antigravity' (custom provider).
+2026-01-24 11:34:30,522 - rotator_library - DEBUG - Handling reasoning parameters: model=antigravity/claude-opus-4.5, reasoning_effort=None
+2026-01-24 11:34:30,523 - rotator_library - DEBUG - Credential priorities for antigravity: P2=1
+2026-01-24 11:34:30,525 - rotator_library - DEBUG - Could not lazy-load tier from /app/oauth_creds/antigravity_oauth_2.json: [Errno 2] No such file or directory: '/app/oauth_creds/antigravity_oauth_2.json'
+2026-01-24 11:34:30,526 - rotator_library - DEBUG - Provider 'antigravity' has custom logic. Delegating call.
+2026-01-24 11:34:30,641 - rotator_library - DEBUG - [Thinking Sanitization] thinking_enabled=True, in_tool_loop=False, turn_has_thinking=False, turn_start_idx=-1
+2026-01-24 11:34:30,641 - rotator_library - DEBUG - [Interleaved Thinking] Injected reminder to user message at index 2
+2026-01-24 11:34:30,642 - rotator_library - DEBUG - [Schema] Preserving property 'pattern' (matches validation keyword name)
+2026-01-24 11:34:30,642 - rotator_library - DEBUG - [Schema] Preserving property 'pattern' (matches validation keyword name)
+2026-01-24 11:34:30,642 - rotator_library - DEBUG - [Schema] Preserving property 'format' (matches validation keyword name)
+2026-01-24 11:34:30,642 - rotator_library - DEBUG - [Schema] Preserving property 'pattern' (matches validation keyword name)
+2026-01-24 11:34:30,643 - rotator_library - DEBUG - [Schema] Preserving property 'pattern' (matches validation keyword name)
+2026-01-24 11:34:30,643 - rotator_library - DEBUG - Starting Antigravity project discovery for credential: /app/oauth_creds/antigravity_oauth_1.json
+2026-01-24 11:34:30,644 - rotator_library - DEBUG - Using cached project ID: hopeful-inkwell-mwz46
+2026-01-24 11:34:30,650 - rotator_library - DEBUG - Handling reasoning parameters: model=antigravity/gemini-3-flash, reasoning_effort=None
+2026-01-24 11:34:30,652 - rotator_library - DEBUG - Credential priorities for antigravity: P2=1
+2026-01-24 11:34:30,653 - rotator_library - DEBUG - Could not lazy-load tier from /app/oauth_creds/antigravity_oauth_2.json: [Errno 2] No such file or directory: '/app/oauth_creds/antigravity_oauth_2.json'
+2026-01-24 11:34:30,654 - rotator_library - DEBUG - Provider 'antigravity' has custom logic. Delegating call.
+2026-01-24 11:34:30,654 - rotator_library - DEBUG - Starting Antigravity project discovery for credential: /app/oauth_creds/antigravity_oauth_1.json
+2026-01-24 11:34:30,654 - rotator_library - DEBUG - Using cached project ID: hopeful-inkwell-mwz46
+2026-01-24 11:34:33,408 - rotator_library - DEBUG - Skipping cost calculation for provider 'antigravity' (custom provider).
+2026-01-24 11:34:36,075 - rotator_library - DEBUG - Skipping cost calculation for provider 'antigravity' (custom provider).
+2026-01-24 11:34:38,581 - rotator_library - DEBUG - Handling reasoning parameters: model=antigravity/claude-opus-4.5, reasoning_effort=None
+2026-01-24 11:34:38,582 - rotator_library - DEBUG - Credential priorities for antigravity: P2=1
+2026-01-24 11:34:38,583 - rotator_library - DEBUG - Could not lazy-load tier from /app/oauth_creds/antigravity_oauth_2.json: [Errno 2] No such file or directory: '/app/oauth_creds/antigravity_oauth_2.json'
+2026-01-24 11:34:38,584 - rotator_library - DEBUG - Provider 'antigravity' has custom logic. Delegating call.
+2026-01-24 11:34:38,585 - rotator_library - DEBUG - [Thinking Sanitization] thinking_enabled=True, in_tool_loop=True, turn_has_thinking=False, turn_start_idx=3
+2026-01-24 11:34:38,585 - rotator_library - DEBUG - [Interleaved Thinking] Injected reminder to user message at index 6
+2026-01-24 11:34:38,586 - rotator_library - DEBUG - [Schema] Preserving property 'pattern' (matches validation keyword name)
+2026-01-24 11:34:38,586 - rotator_library - DEBUG - [Schema] Preserving property 'pattern' (matches validation keyword name)
+2026-01-24 11:34:38,586 - rotator_library - DEBUG - [Schema] Preserving property 'format' (matches validation keyword name)
+2026-01-24 11:34:38,586 - rotator_library - DEBUG - [Schema] Preserving property 'pattern' (matches validation keyword name)
+2026-01-24 11:34:38,587 - rotator_library - DEBUG - [Schema] Preserving property 'pattern' (matches validation keyword name)
+2026-01-24 11:34:38,587 - rotator_library - DEBUG - Starting Antigravity project discovery for credential: /app/oauth_creds/antigravity_oauth_1.json
+2026-01-24 11:34:38,588 - rotator_library - DEBUG - Using cached project ID: hopeful-inkwell-mwz46
+2026-01-24 11:34:44,008 - rotator_library - DEBUG - Skipping cost calculation for provider 'antigravity' (custom provider).
+2026-01-24 11:35:14,577 - rotator_library - DEBUG - Refreshing ANTIGRAVITY OAuth token for 'antigravity_oauth_1.json' (forced: False)...
+2026-01-24 11:35:15,337 - rotator_library - DEBUG - Token validation successful for 'antigravity_oauth_1.json'
+2026-01-24 11:35:15,448 - rotator_library - DEBUG - Saved updated ANTIGRAVITY OAuth credentials to '/app/oauth_creds/antigravity_oauth_1.json'.
+2026-01-24 11:35:15,449 - rotator_library - DEBUG - Successfully refreshed ANTIGRAVITY OAuth token for 'antigravity_oauth_1.json'.
+2026-01-24 11:35:21,124 - rotator_library - DEBUG - Refreshing quota baselines for 1 recently active credentials
+2026-01-24 11:35:21,806 - rotator_library - DEBUG - Updated quota baseline for antigravity_oauth_1.json model=antigravity/gemini-2.5-flash: remaining=100.00%, synced_request_count=1
+2026-01-24 11:35:21,807 - rotator_library - DEBUG - Updated quota baseline for antigravity_oauth_1.json model=antigravity/gemini-2.5-flash-lite: remaining=100.00%, synced_request_count=0
+2026-01-24 11:35:21,809 - rotator_library - DEBUG - Updated quota baseline for antigravity_oauth_1.json model=antigravity/claude-sonnet-4.5: remaining=100.00%, synced_request_count=3
+2026-01-24 11:35:21,810 - rotator_library - DEBUG - Updated quota baseline for antigravity_oauth_1.json model=antigravity/gemini-3-flash: remaining=100.00%, synced_request_count=3
+2026-01-24 11:35:21,812 - rotator_library - DEBUG - Updated quota baseline for antigravity_oauth_1.json model=antigravity/gemini-3-pro-preview: remaining=100.00%, synced_request_count=0
+2026-01-24 11:35:21,813 - rotator_library - DEBUG - Updated quota baseline for antigravity_oauth_1.json model=antigravity/claude-opus-4.5: remaining=100.00%, synced_request_count=3
+2026-01-24 11:35:21,814 - rotator_library - DEBUG - Antigravity quota baseline refresh: no cooldowns needed
+2026-01-24 11:35:21,815 - rotator_library - DEBUG - antigravity quota refresh: updated 6 model baselines
+2026-01-24 11:35:21,815 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 11:38:54,075 - rotator_library - DEBUG - Handling reasoning parameters: model=antigravity/claude-opus-4.5, reasoning_effort=None
+2026-01-24 11:38:54,076 - rotator_library - DEBUG - Credential priorities for antigravity: P2=1
+2026-01-24 11:38:54,077 - rotator_library - DEBUG - Could not lazy-load tier from /app/oauth_creds/antigravity_oauth_2.json: [Errno 2] No such file or directory: '/app/oauth_creds/antigravity_oauth_2.json'
+2026-01-24 11:38:54,080 - rotator_library - DEBUG - Provider 'antigravity' has custom logic. Delegating call.
+2026-01-24 11:38:54,081 - rotator_library - DEBUG - [Thinking Sanitization] thinking_enabled=True, in_tool_loop=False, turn_has_thinking=False, turn_start_idx=-1
+2026-01-24 11:38:54,081 - rotator_library - DEBUG - [Interleaved Thinking] Injected reminder to user message at index 6
+2026-01-24 11:38:54,082 - rotator_library - DEBUG - [Schema] Preserving property 'pattern' (matches validation keyword name)
+2026-01-24 11:38:54,087 - rotator_library - DEBUG - [Schema] Preserving property 'pattern' (matches validation keyword name)
+2026-01-24 11:38:54,087 - rotator_library - DEBUG - [Schema] Preserving property 'format' (matches validation keyword name)
+2026-01-24 11:38:54,088 - rotator_library - DEBUG - [Schema] Preserving property 'pattern' (matches validation keyword name)
+2026-01-24 11:38:54,088 - rotator_library - DEBUG - [Schema] Preserving property 'pattern' (matches validation keyword name)
+2026-01-24 11:38:54,088 - rotator_library - DEBUG - Starting Antigravity project discovery for credential: /app/oauth_creds/antigravity_oauth_1.json
+2026-01-24 11:38:54,089 - rotator_library - DEBUG - Using cached project ID: hopeful-inkwell-mwz46
+2026-01-24 11:38:54,148 - rotator_library - DEBUG - Handling reasoning parameters: model=antigravity/gemini-3-flash, reasoning_effort=None
+2026-01-24 11:38:54,150 - rotator_library - DEBUG - Credential priorities for antigravity: P2=1
+2026-01-24 11:38:54,151 - rotator_library - DEBUG - Could not lazy-load tier from /app/oauth_creds/antigravity_oauth_2.json: [Errno 2] No such file or directory: '/app/oauth_creds/antigravity_oauth_2.json'
+2026-01-24 11:38:54,152 - rotator_library - DEBUG - Provider 'antigravity' has custom logic. Delegating call.
+2026-01-24 11:38:54,153 - rotator_library - DEBUG - Starting Antigravity project discovery for credential: /app/oauth_creds/antigravity_oauth_1.json
+2026-01-24 11:38:54,153 - rotator_library - DEBUG - Using cached project ID: hopeful-inkwell-mwz46
+2026-01-24 11:38:57,000 - rotator_library - DEBUG - Skipping cost calculation for provider 'antigravity' (custom provider).
+2026-01-24 11:38:58,818 - rotator_library - DEBUG - Skipping cost calculation for provider 'antigravity' (custom provider).
+2026-01-24 11:39:23,427 - rotator_library - DEBUG - Handling reasoning parameters: model=antigravity/gemini-3-flash, reasoning_effort=None
+2026-01-24 11:39:23,428 - rotator_library - DEBUG - Credential priorities for antigravity: P2=1
+2026-01-24 11:39:23,429 - rotator_library - DEBUG - Could not lazy-load tier from /app/oauth_creds/antigravity_oauth_2.json: [Errno 2] No such file or directory: '/app/oauth_creds/antigravity_oauth_2.json'
+2026-01-24 11:39:23,430 - rotator_library - DEBUG - Provider 'antigravity' has custom logic. Delegating call.
+2026-01-24 11:39:23,431 - rotator_library - DEBUG - Missing thoughtSignature for first func call toolu_vrtx_01AxDghXwH4PMXNES8HtfNv9, using bypass
+2026-01-24 11:39:23,431 - rotator_library - DEBUG - [Schema] Preserving property 'pattern' (matches validation keyword name)
+2026-01-24 11:39:23,431 - rotator_library - DEBUG - [Schema] Preserving property 'pattern' (matches validation keyword name)
+2026-01-24 11:39:23,431 - rotator_library - DEBUG - [Schema] Preserving property 'format' (matches validation keyword name)
+2026-01-24 11:39:23,432 - rotator_library - DEBUG - [Schema] Preserving property 'pattern' (matches validation keyword name)
+2026-01-24 11:39:23,432 - rotator_library - DEBUG - [Schema] Preserving property 'pattern' (matches validation keyword name)
+2026-01-24 11:39:23,440 - rotator_library - DEBUG - Starting Antigravity project discovery for credential: /app/oauth_creds/antigravity_oauth_1.json
+2026-01-24 11:39:23,441 - rotator_library - DEBUG - Using cached project ID: hopeful-inkwell-mwz46
+2026-01-24 11:39:23,450 - rotator_library - DEBUG - Handling reasoning parameters: model=antigravity/gemini-3-flash, reasoning_effort=None
+2026-01-24 11:39:23,451 - rotator_library - DEBUG - Credential priorities for antigravity: P2=1
+2026-01-24 11:39:23,452 - rotator_library - DEBUG - Could not lazy-load tier from /app/oauth_creds/antigravity_oauth_2.json: [Errno 2] No such file or directory: '/app/oauth_creds/antigravity_oauth_2.json'
+2026-01-24 11:39:23,453 - rotator_library - DEBUG - Provider 'antigravity' has custom logic. Delegating call.
+2026-01-24 11:39:23,454 - rotator_library - DEBUG - Starting Antigravity project discovery for credential: /app/oauth_creds/antigravity_oauth_1.json
+2026-01-24 11:39:23,454 - rotator_library - DEBUG - Using cached project ID: hopeful-inkwell-mwz46
+2026-01-24 11:39:25,863 - rotator_library - DEBUG - Skipping cost calculation for provider 'antigravity' (custom provider).
+2026-01-24 11:39:27,283 - rotator_library - DEBUG - Skipping cost calculation for provider 'antigravity' (custom provider).
+2026-01-24 11:40:21,816 - rotator_library - DEBUG - Refreshing quota baselines for 1 recently active credentials
+2026-01-24 11:40:22,455 - rotator_library - DEBUG - Updated quota baseline for antigravity_oauth_1.json model=antigravity/gemini-3-pro-preview: remaining=100.00%, synced_request_count=0
+2026-01-24 11:40:22,456 - rotator_library - DEBUG - Updated quota baseline for antigravity_oauth_1.json model=antigravity/gemini-2.5-flash-lite: remaining=100.00%, synced_request_count=0
+2026-01-24 11:40:22,475 - rotator_library - DEBUG - Updated quota baseline for antigravity_oauth_1.json model=antigravity/gemini-2.5-flash: remaining=100.00%, synced_request_count=1
+2026-01-24 11:40:22,477 - rotator_library - DEBUG - Updated quota baseline for antigravity_oauth_1.json model=antigravity/gemini-3-flash: remaining=100.00%, synced_request_count=6
+2026-01-24 11:40:22,479 - rotator_library - DEBUG - Updated quota baseline for antigravity_oauth_1.json model=antigravity/claude-sonnet-4.5: remaining=100.00%, synced_request_count=4
+2026-01-24 11:40:22,480 - rotator_library - DEBUG - Updated quota baseline for antigravity_oauth_1.json model=antigravity/claude-opus-4.5: remaining=100.00%, synced_request_count=4
+2026-01-24 11:40:22,612 - rotator_library - DEBUG - Antigravity quota baseline refresh: no cooldowns needed
+2026-01-24 11:40:22,613 - rotator_library - DEBUG - antigravity quota refresh: updated 6 model baselines
+2026-01-24 11:40:22,613 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 11:45:22,613 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 11:45:22,613 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 11:50:22,617 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 11:50:22,617 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 11:55:22,618 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 11:55:22,620 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 12:00:22,621 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 12:00:22,621 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 12:05:14,583 - rotator_library - DEBUG - Refreshing ANTIGRAVITY OAuth token for 'antigravity_oauth_1.json' (forced: False)...
+2026-01-24 12:05:15,498 - rotator_library - DEBUG - Token validation successful for 'antigravity_oauth_1.json'
+2026-01-24 12:05:15,500 - rotator_library - DEBUG - Saved updated ANTIGRAVITY OAuth credentials to '/app/oauth_creds/antigravity_oauth_1.json'.
+2026-01-24 12:05:15,500 - rotator_library - DEBUG - Successfully refreshed ANTIGRAVITY OAuth token for 'antigravity_oauth_1.json'.
+2026-01-24 12:05:22,621 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 12:05:22,622 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 12:10:22,622 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 12:10:22,623 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 12:15:22,624 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 12:15:22,624 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 12:20:22,625 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 12:20:22,625 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 12:25:22,626 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 12:25:22,627 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 12:30:22,627 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 12:30:22,627 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 12:35:14,587 - rotator_library - DEBUG - Refreshing ANTIGRAVITY OAuth token for 'antigravity_oauth_1.json' (forced: False)...
+2026-01-24 12:35:15,520 - rotator_library - DEBUG - Token validation successful for 'antigravity_oauth_1.json'
+2026-01-24 12:35:15,521 - rotator_library - DEBUG - Saved updated ANTIGRAVITY OAuth credentials to '/app/oauth_creds/antigravity_oauth_1.json'.
+2026-01-24 12:35:15,522 - rotator_library - DEBUG - Successfully refreshed ANTIGRAVITY OAuth token for 'antigravity_oauth_1.json'.
+2026-01-24 12:35:22,628 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 12:35:22,628 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 12:40:22,630 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 12:40:22,630 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 12:45:22,631 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 12:45:22,632 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 12:50:22,632 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 12:50:22,632 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 12:55:22,634 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 12:55:22,636 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 13:00:22,637 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 13:00:22,638 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 13:05:14,590 - rotator_library - DEBUG - Refreshing ANTIGRAVITY OAuth token for 'antigravity_oauth_1.json' (forced: False)...
+2026-01-24 13:05:15,435 - rotator_library - DEBUG - Token validation successful for 'antigravity_oauth_1.json'
+2026-01-24 13:05:15,437 - rotator_library - DEBUG - Saved updated ANTIGRAVITY OAuth credentials to '/app/oauth_creds/antigravity_oauth_1.json'.
+2026-01-24 13:05:15,437 - rotator_library - DEBUG - Successfully refreshed ANTIGRAVITY OAuth token for 'antigravity_oauth_1.json'.
+2026-01-24 13:05:22,638 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 13:05:22,639 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 13:10:22,640 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 13:10:22,640 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 13:15:22,641 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 13:15:22,642 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 13:20:27,937 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 13:20:27,937 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 13:25:27,941 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 13:25:27,941 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 13:30:27,941 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 13:30:27,941 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 13:35:19,887 - rotator_library - DEBUG - Refreshing ANTIGRAVITY OAuth token for 'antigravity_oauth_1.json' (forced: False)...
+2026-01-24 13:35:20,806 - rotator_library - DEBUG - Token validation successful for 'antigravity_oauth_1.json'
+2026-01-24 13:35:20,807 - rotator_library - DEBUG - Saved updated ANTIGRAVITY OAuth credentials to '/app/oauth_creds/antigravity_oauth_1.json'.
+2026-01-24 13:35:20,807 - rotator_library - DEBUG - Successfully refreshed ANTIGRAVITY OAuth token for 'antigravity_oauth_1.json'.
+2026-01-24 13:35:27,942 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 13:35:27,942 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 13:40:27,943 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 13:40:27,944 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 13:45:27,945 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 13:45:27,946 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 13:50:27,946 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 13:50:27,951 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 13:55:27,951 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 13:55:27,951 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 14:00:27,952 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 14:00:27,952 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 14:05:19,891 - rotator_library - DEBUG - Refreshing ANTIGRAVITY OAuth token for 'antigravity_oauth_1.json' (forced: False)...
+2026-01-24 14:05:20,652 - rotator_library - DEBUG - Token validation successful for 'antigravity_oauth_1.json'
+2026-01-24 14:05:20,655 - rotator_library - DEBUG - Saved updated ANTIGRAVITY OAuth credentials to '/app/oauth_creds/antigravity_oauth_1.json'.
+2026-01-24 14:05:20,655 - rotator_library - DEBUG - Successfully refreshed ANTIGRAVITY OAuth token for 'antigravity_oauth_1.json'.
+2026-01-24 14:05:27,952 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 14:05:27,952 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 14:10:27,953 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 14:10:27,953 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 14:15:27,954 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 14:15:27,957 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 14:20:27,958 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 14:20:27,959 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 14:25:27,960 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 14:25:27,960 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 14:30:27,961 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 14:30:27,961 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 14:35:19,895 - rotator_library - DEBUG - Refreshing ANTIGRAVITY OAuth token for 'antigravity_oauth_1.json' (forced: False)...
+2026-01-24 14:35:20,722 - rotator_library - DEBUG - Token validation successful for 'antigravity_oauth_1.json'
+2026-01-24 14:35:20,723 - rotator_library - DEBUG - Saved updated ANTIGRAVITY OAuth credentials to '/app/oauth_creds/antigravity_oauth_1.json'.
+2026-01-24 14:35:20,723 - rotator_library - DEBUG - Successfully refreshed ANTIGRAVITY OAuth token for 'antigravity_oauth_1.json'.
+2026-01-24 14:35:27,962 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 14:35:27,964 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 14:40:27,964 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 14:40:27,965 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 14:45:27,966 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 14:45:27,966 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 14:50:27,967 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 14:50:27,970 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 14:55:27,971 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 14:55:27,971 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 15:00:27,972 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 15:00:27,972 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 15:05:19,897 - rotator_library - DEBUG - Refreshing ANTIGRAVITY OAuth token for 'antigravity_oauth_1.json' (forced: False)...
+2026-01-24 15:05:20,696 - rotator_library - DEBUG - Token validation successful for 'antigravity_oauth_1.json'
+2026-01-24 15:05:20,697 - rotator_library - DEBUG - Saved updated ANTIGRAVITY OAuth credentials to '/app/oauth_creds/antigravity_oauth_1.json'.
+2026-01-24 15:05:20,697 - rotator_library - DEBUG - Successfully refreshed ANTIGRAVITY OAuth token for 'antigravity_oauth_1.json'.
+2026-01-24 15:05:27,973 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 15:05:27,973 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 15:10:27,974 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 15:10:27,974 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 15:15:27,975 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 15:15:27,977 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 15:20:27,977 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 15:20:27,977 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 15:25:27,978 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 15:25:27,978 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 15:30:27,979 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 15:30:27,980 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 15:35:19,899 - rotator_library - DEBUG - Refreshing ANTIGRAVITY OAuth token for 'antigravity_oauth_1.json' (forced: False)...
+2026-01-24 15:35:20,685 - rotator_library - DEBUG - Token validation successful for 'antigravity_oauth_1.json'
+2026-01-24 15:35:20,686 - rotator_library - DEBUG - Saved updated ANTIGRAVITY OAuth credentials to '/app/oauth_creds/antigravity_oauth_1.json'.
+2026-01-24 15:35:20,686 - rotator_library - DEBUG - Successfully refreshed ANTIGRAVITY OAuth token for 'antigravity_oauth_1.json'.
+2026-01-24 15:35:27,982 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 15:35:27,982 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 15:40:27,983 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 15:40:27,983 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 15:45:27,984 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 15:45:27,984 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 15:50:27,985 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 15:50:27,987 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 15:55:27,988 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 15:55:27,988 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 16:00:27,989 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 16:00:27,989 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 16:05:19,902 - rotator_library - DEBUG - Refreshing ANTIGRAVITY OAuth token for 'antigravity_oauth_1.json' (forced: False)...
+2026-01-24 16:05:20,610 - rotator_library - DEBUG - Token validation successful for 'antigravity_oauth_1.json'
+2026-01-24 16:05:20,613 - rotator_library - DEBUG - Saved updated ANTIGRAVITY OAuth credentials to '/app/oauth_creds/antigravity_oauth_1.json'.
+2026-01-24 16:05:20,613 - rotator_library - DEBUG - Successfully refreshed ANTIGRAVITY OAuth token for 'antigravity_oauth_1.json'.
+2026-01-24 16:05:27,990 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 16:05:27,990 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 16:10:27,991 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 16:10:27,992 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 16:15:27,993 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 16:15:27,997 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 16:20:27,999 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 16:20:27,999 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 16:25:28,000 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 16:25:28,000 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 16:30:28,001 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 16:30:28,001 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 16:35:19,905 - rotator_library - DEBUG - Refreshing ANTIGRAVITY OAuth token for 'antigravity_oauth_1.json' (forced: False)...
+2026-01-24 16:35:20,764 - rotator_library - DEBUG - Token validation successful for 'antigravity_oauth_1.json'
+2026-01-24 16:35:20,765 - rotator_library - DEBUG - Saved updated ANTIGRAVITY OAuth credentials to '/app/oauth_creds/antigravity_oauth_1.json'.
+2026-01-24 16:35:20,766 - rotator_library - DEBUG - Successfully refreshed ANTIGRAVITY OAuth token for 'antigravity_oauth_1.json'.
+2026-01-24 16:35:28,003 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 16:35:28,004 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 16:40:28,005 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 16:40:28,005 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 16:45:28,006 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 16:45:28,006 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 16:50:28,007 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 16:50:28,008 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 16:55:28,008 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 16:55:28,008 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 17:00:28,009 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 17:00:28,009 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 17:05:19,906 - rotator_library - DEBUG - Refreshing ANTIGRAVITY OAuth token for 'antigravity_oauth_1.json' (forced: False)...
+2026-01-24 17:05:28,010 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 17:05:28,014 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete
+2026-01-24 17:06:04,909 - rotator_library - DEBUG - Refreshing ANTIGRAVITY OAuth token for 'antigravity_oauth_1.json' (forced: False)...
+2026-01-24 17:06:49,911 - rotator_library - DEBUG - Refreshing ANTIGRAVITY OAuth token for 'antigravity_oauth_1.json' (forced: False)...
+2026-01-24 17:10:28,015 - rotator_library - DEBUG - No recently active credentials to refresh quota baselines
+2026-01-24 17:10:28,016 - rotator_library - DEBUG - antigravity antigravity_quota_refresh: periodic run complete

requirements.txt

@@ -0,0 +1,27 @@
+# FastAPI framework for building the proxy server
+fastapi
+# ASGI server for running the FastAPI application
+uvicorn
+# For loading environment variables from a .env file
+python-dotenv
+
+# Installs the local rotator_library in editable mode
+-e src/rotator_library
+
+# A library for calling LLM APIs with a consistent format
+litellm
+
+filelock
+httpx
+aiofiles
+aiohttp
+
+colorlog
+
+rich
+
+# GUI for model filter configuration
+customtkinter
+
+# For building the executable
+pyinstaller

src/batch_auth.py

@@ -0,0 +1,40 @@
+# SPDX-License-Identifier: LGPL-3.0-only
+# Copyright (c) 2026 Mirrowel
+
+import asyncio
+import sys
+import argparse
+from pathlib import Path
+
+# Add the 'src' directory to the Python path
+sys.path.append(str(Path(__file__).resolve().parent))
+
+from rotator_library import provider_factory
+
+async def main():
+    parser = argparse.ArgumentParser(description="Batch authorize multiple Google OAuth accounts.")
+    parser.add_argument("emails", nargs="+", help="List of Gmail addresses to authorize.")
+    parser.add_argument("--provider", default="antigravity", help="Provider to authorize (default: antigravity).")
+    args = parser.parse_args()
+
+    auth_class = provider_factory.get_provider_auth_class(args.provider)
+    auth_instance = auth_class()
+
+    print(f"🚀 Starting batch authorization for {len(args.emails)} accounts on {args.provider}...")
+
+    for email in args.emails:
+        print(f"\n🔑 Setting up: {email}")
+        result = await auth_instance.setup_credential(login_hint=email)
+        
+        if result.success:
+            print(f"✅ Success! Saved to: {Path(result.file_path).name}")
+            if result.is_update:
+                print(f"ℹ️  Updated existing credential for {result.email}")
+        else:
+            print(f"❌ Failed: {result.error}")
+
+    print("\n✨ Batch authorization complete!")
+    print("👉 Now run 'python -m rotator_library.credential_tool' and choose 'Export to .env' to get your tokens.")
+
+if __name__ == "__main__":
+    asyncio.run(main())

src/proxy_app/LICENSE

@@ -0,0 +1,21 @@
+MIT License
+
+Copyright (c) 2025 Mirrowel
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.

src/proxy_app/__init__.py

@@ -0,0 +1,3 @@
+# SPDX-License-Identifier: MIT
+# Copyright (c) 2026 Mirrowel
+

src/proxy_app/batch_manager.py

@@ -0,0 +1,84 @@
+# SPDX-License-Identifier: MIT
+# Copyright (c) 2026 Mirrowel
+
+import asyncio
+from typing import List, Dict, Any, Tuple
+import time
+from rotator_library import RotatingClient
+
+class EmbeddingBatcher:
+    def __init__(self, client: RotatingClient, batch_size: int = 64, timeout: float = 0.1):
+        self.client = client
+        self.batch_size = batch_size
+        self.timeout = timeout
+        self.queue = asyncio.Queue()
+        self.worker_task = asyncio.create_task(self._batch_worker())
+
+    async def add_request(self, request_data: Dict[str, Any]) -> Any:
+        future = asyncio.Future()
+        await self.queue.put((request_data, future))
+        return await future
+
+    async def _batch_worker(self):
+        while True:
+            batch, futures = await self._gather_batch()
+            if not batch:
+                continue
+
+            try:
+                # Assume all requests in a batch use the same model and other settings
+                model = batch[0]["model"]
+                inputs = [item["input"][0] for item in batch] # Extract single string input
+
+                batched_request = {
+                    "model": model,
+                    "input": inputs
+                }
+                
+                # Pass through any other relevant parameters from the first request
+                for key in ["input_type", "dimensions", "user"]:
+                    if key in batch[0]:
+                        batched_request[key] = batch[0][key]
+
+                response = await self.client.aembedding(**batched_request)
+                
+                # Distribute results back to the original requesters
+                for i, future in enumerate(futures):
+                    # Create a new response object for each item in the batch
+                    single_response_data = {
+                        "object": response.object,
+                        "model": response.model,
+                        "data": [response.data[i]],
+                        "usage": response.usage # Usage is for the whole batch
+                    }
+                    future.set_result(single_response_data)
+
+            except Exception as e:
+                for future in futures:
+                    future.set_exception(e)
+
+    async def _gather_batch(self) -> Tuple[List[Dict[str, Any]], List[asyncio.Future]]:
+        batch = []
+        futures = []
+        start_time = time.time()
+
+        while len(batch) < self.batch_size and (time.time() - start_time) < self.timeout:
+            try:
+                # Wait for an item with a timeout
+                timeout = self.timeout - (time.time() - start_time)
+                if timeout <= 0:
+                    break
+                request, future = await asyncio.wait_for(self.queue.get(), timeout=timeout)
+                batch.append(request)
+                futures.append(future)
+            except asyncio.TimeoutError:
+                break
+        
+        return batch, futures
+
+    async def stop(self):
+        self.worker_task.cancel()
+        try:
+            await self.worker_task
+        except asyncio.CancelledError:
+            pass

src/proxy_app/build.py

@@ -0,0 +1,95 @@
+# SPDX-License-Identifier: MIT
+# Copyright (c) 2026 Mirrowel
+
+import os
+import sys
+import platform
+import subprocess
+
+
+def get_providers():
+    """
+    Scans the 'src/rotator_library/providers' directory to find all provider modules.
+    Returns a list of hidden import arguments for PyInstaller.
+    """
+    hidden_imports = []
+    # Get the absolute path to the directory containing this script
+    script_dir = os.path.dirname(os.path.abspath(__file__))
+    # Construct the path to the providers directory relative to this script's location
+    providers_path = os.path.join(script_dir, "..", "rotator_library", "providers")
+
+    if not os.path.isdir(providers_path):
+        print(f"Error: Directory not found at '{os.path.abspath(providers_path)}'")
+        return []
+
+    for filename in os.listdir(providers_path):
+        if filename.endswith("_provider.py") and filename != "__init__.py":
+            module_name = f"rotator_library.providers.{filename[:-3]}"
+            hidden_imports.append(f"--hidden-import={module_name}")
+    return hidden_imports
+
+
+def main():
+    """
+    Constructs and runs the PyInstaller command to build the executable.
+    """
+    # Base PyInstaller command with optimizations
+    command = [
+        sys.executable,
+        "-m",
+        "PyInstaller",
+        "--onefile",
+        "--name",
+        "proxy_app",
+        "--paths",
+        "../",
+        "--paths",
+        ".",
+        # Core imports
+        "--hidden-import=rotator_library",
+        "--hidden-import=tiktoken_ext.openai_public",
+        "--hidden-import=tiktoken_ext",
+        "--collect-data",
+        "litellm",
+        # Optimization: Exclude unused heavy modules
+        "--exclude-module=matplotlib",
+        "--exclude-module=IPython",
+        "--exclude-module=jupyter",
+        "--exclude-module=notebook",
+        "--exclude-module=PIL.ImageTk",
+        # Optimization: Enable UPX compression (if available)
+        "--upx-dir=upx"
+        if platform.system() != "Darwin"
+        else "--noupx",  # macOS has issues with UPX
+        # Optimization: Strip debug symbols (smaller binary)
+        "--strip"
+        if platform.system() != "Windows"
+        else "--console",  # Windows gets clean console
+    ]
+
+    # Add hidden imports for providers
+    provider_imports = get_providers()
+    if not provider_imports:
+        print(
+            "Warning: No providers found. The build might not include any LLM providers."
+        )
+    command.extend(provider_imports)
+
+    # Add the main script
+    command.append("main.py")
+
+    # Execute the command
+    print(f"Running command: {' '.join(command)}")
+    try:
+        # Run PyInstaller from the script's directory to ensure relative paths are correct
+        script_dir = os.path.dirname(os.path.abspath(__file__))
+        subprocess.run(command, check=True, cwd=script_dir)
+        print("Build successful!")
+    except subprocess.CalledProcessError as e:
+        print(f"Build failed with error: {e}")
+    except FileNotFoundError:
+        print("Error: PyInstaller is not installed or not in the system's PATH.")
+
+
+if __name__ == "__main__":
+    main()

src/proxy_app/detailed_logger.py

@@ -0,0 +1,187 @@
+# SPDX-License-Identifier: MIT
+# Copyright (c) 2026 Mirrowel
+
+# src/proxy_app/detailed_logger.py
+"""
+Raw I/O Logger for the Proxy Layer.
+
+This logger captures the UNMODIFIED HTTP request and response at the proxy boundary.
+It is disabled by default and should only be enabled for debugging the proxy itself.
+
+Use this when you need to:
+- Verify that requests/responses are not being corrupted
+- Debug HTTP-level issues between the client and proxy
+- Capture exact payloads as received/sent by the proxy
+
+For normal request/response logging with provider correlation, use the
+TransactionLogger in the rotator_library instead (enabled via --enable-request-logging).
+
+Directory structure:
+    logs/raw_io/{YYYYMMDD_HHMMSS}_{request_id}/
+        request.json           # Unmodified incoming HTTP request
+        streaming_chunks.jsonl # If streaming mode
+        final_response.json    # Unmodified outgoing HTTP response
+        metadata.json          # Summary metadata
+"""
+
+import json
+import time
+import uuid
+from datetime import datetime
+from pathlib import Path
+from typing import Any, Dict, Optional
+import logging
+
+from rotator_library.utils.resilient_io import (
+    safe_write_json,
+    safe_log_write,
+    safe_mkdir,
+)
+from rotator_library.utils.paths import get_logs_dir
+
+
+def _get_raw_io_logs_dir() -> Path:
+    """Get the raw I/O logs directory, creating it if needed."""
+    logs_dir = get_logs_dir()
+    raw_io_dir = logs_dir / "raw_io"
+    raw_io_dir.mkdir(parents=True, exist_ok=True)
+    return raw_io_dir
+
+
+class RawIOLogger:
+    """
+    Logs raw HTTP request/response at the proxy boundary.
+
+    This captures the EXACT data as received from and sent to the client,
+    without any transformations. Useful for debugging the proxy itself.
+
+    DISABLED by default. Enable with --enable-raw-logging flag.
+
+    Uses fire-and-forget logging - if disk writes fail, logs are dropped (not buffered)
+    to prevent memory issues, especially with streaming responses.
+    """
+
+    def __init__(self):
+        """
+        Initializes the logger for a single request, creating a unique directory
+        to store all related log files.
+        """
+        self.start_time = time.time()
+        self.request_id = str(uuid.uuid4())
+        timestamp = datetime.now().strftime("%Y%m%d_%H%M%S")
+        self.log_dir = _get_raw_io_logs_dir() / f"{timestamp}_{self.request_id}"
+        self.streaming = False
+        self._dir_available = safe_mkdir(self.log_dir, logging)
+
+    def _write_json(self, filename: str, data: Dict[str, Any]):
+        """Helper to write data to a JSON file in the log directory."""
+        if not self._dir_available:
+            # Try to create directory again in case it was recreated
+            self._dir_available = safe_mkdir(self.log_dir, logging)
+            if not self._dir_available:
+                return
+
+        safe_write_json(
+            self.log_dir / filename,
+            data,
+            logging,
+            atomic=False,
+            indent=4,
+            ensure_ascii=False,
+        )
+
+    def log_request(self, headers: Dict[str, Any], body: Dict[str, Any]):
+        """Logs the raw incoming request details."""
+        self.streaming = body.get("stream", False)
+        request_data = {
+            "request_id": self.request_id,
+            "timestamp_utc": datetime.utcnow().isoformat(),
+            "headers": dict(headers),
+            "body": body,
+        }
+        self._write_json("request.json", request_data)
+
+    def log_stream_chunk(self, chunk: Dict[str, Any]):
+        """Logs an individual chunk from a streaming response to a JSON Lines file."""
+        if not self._dir_available:
+            return
+
+        log_entry = {"timestamp_utc": datetime.utcnow().isoformat(), "chunk": chunk}
+        content = json.dumps(log_entry, ensure_ascii=False) + "\n"
+        safe_log_write(self.log_dir / "streaming_chunks.jsonl", content, logging)
+
+    def log_final_response(
+        self, status_code: int, headers: Optional[Dict[str, Any]], body: Dict[str, Any]
+    ):
+        """Logs the raw outgoing response."""
+        end_time = time.time()
+        duration_ms = (end_time - self.start_time) * 1000
+
+        response_data = {
+            "request_id": self.request_id,
+            "timestamp_utc": datetime.utcnow().isoformat(),
+            "status_code": status_code,
+            "duration_ms": round(duration_ms),
+            "headers": dict(headers) if headers else None,
+            "body": body,
+        }
+        self._write_json("final_response.json", response_data)
+        self._log_metadata(response_data)
+
+    def _extract_reasoning(self, response_body: Dict[str, Any]) -> Optional[str]:
+        """Recursively searches for and extracts 'reasoning' fields from the response body."""
+        if not isinstance(response_body, dict):
+            return None
+
+        if "reasoning" in response_body:
+            return response_body["reasoning"]
+
+        if "choices" in response_body and response_body["choices"]:
+            message = response_body["choices"][0].get("message", {})
+            if "reasoning" in message:
+                return message["reasoning"]
+            if "reasoning_content" in message:
+                return message["reasoning_content"]
+
+        return None
+
+    def _log_metadata(self, response_data: Dict[str, Any]):
+        """Logs a summary of the transaction for quick analysis."""
+        usage = response_data.get("body", {}).get("usage") or {}
+        model = response_data.get("body", {}).get("model", "N/A")
+        finish_reason = "N/A"
+        if (
+            "choices" in response_data.get("body", {})
+            and response_data["body"]["choices"]
+        ):
+            finish_reason = response_data["body"]["choices"][0].get(
+                "finish_reason", "N/A"
+            )
+
+        metadata = {
+            "request_id": self.request_id,
+            "timestamp_utc": response_data["timestamp_utc"],
+            "duration_ms": response_data["duration_ms"],
+            "status_code": response_data["status_code"],
+            "model": model,
+            "streaming": self.streaming,
+            "usage": {
+                "prompt_tokens": usage.get("prompt_tokens"),
+                "completion_tokens": usage.get("completion_tokens"),
+                "total_tokens": usage.get("total_tokens"),
+            },
+            "finish_reason": finish_reason,
+            "reasoning_found": False,
+            "reasoning_content": None,
+        }
+
+        reasoning = self._extract_reasoning(response_data.get("body", {}))
+        if reasoning:
+            metadata["reasoning_found"] = True
+            metadata["reasoning_content"] = reasoning
+
+        self._write_json("metadata.json", metadata)
+
+
+# Backward compatibility alias
+DetailedLogger = RawIOLogger

src/proxy_app/launcher_tui.py

@@ -0,0 +1,1084 @@
+# SPDX-License-Identifier: MIT
+# Copyright (c) 2026 Mirrowel
+
+"""
+Interactive TUI launcher for the LLM API Key Proxy.
+Provides a beautiful Rich-based interface for configuration and execution.
+"""
+
+import json
+import os
+import sys
+from pathlib import Path
+from rich.console import Console
+from rich.prompt import IntPrompt, Prompt
+from rich.panel import Panel
+from rich.text import Text
+from dotenv import load_dotenv, set_key
+
+console = Console()
+
+
+def _get_env_file() -> Path:
+    """
+    Get .env file path (lightweight - no heavy imports).
+
+    Returns:
+        Path to .env file - EXE directory if frozen, else current working directory
+    """
+    if getattr(sys, "frozen", False):
+        # Running as PyInstaller EXE - use EXE's directory
+        return Path(sys.executable).parent / ".env"
+    # Running as script - use current working directory
+    return Path.cwd() / ".env"
+
+
+def clear_screen(subtitle: str = ""):
+    """
+    Cross-platform terminal clear with optional header.
+
+    Uses native OS commands instead of ANSI escape sequences:
+    - Windows (conhost & Windows Terminal): cls
+    - Unix-like systems (Linux, Mac): clear
+
+    Args:
+        subtitle: If provided, displays a header panel with this subtitle.
+                  If empty/None, just clears the screen.
+    """
+    os.system("cls" if os.name == "nt" else "clear")
+    if subtitle:
+        console.print(
+            Panel(
+                f"[bold cyan]{subtitle}[/bold cyan]",
+                title="--- API Key Proxy ---",
+            )
+        )
+
+
+class LauncherConfig:
+    """Manages launcher_config.json (host, port, logging only)"""
+
+    def __init__(self, config_path: Path = Path("launcher_config.json")):
+        self.config_path = config_path
+        self.defaults = {
+            "host": "127.0.0.1",
+            "port": 8000,
+            "enable_request_logging": False,
+            "enable_raw_logging": False,
+        }
+        self.config = self.load()
+
+    def load(self) -> dict:
+        """Load config from file or create with defaults."""
+        if self.config_path.exists():
+            try:
+                with open(self.config_path, "r") as f:
+                    config = json.load(f)
+                # Merge with defaults for any missing keys
+                for key, value in self.defaults.items():
+                    if key not in config:
+                        config[key] = value
+                return config
+            except (json.JSONDecodeError, IOError):
+                return self.defaults.copy()
+        return self.defaults.copy()
+
+    def save(self):
+        """Save current config to file."""
+        import datetime
+
+        self.config["last_updated"] = datetime.datetime.now().isoformat()
+        try:
+            with open(self.config_path, "w") as f:
+                json.dump(self.config, f, indent=2)
+        except IOError as e:
+            console.print(f"[red]Error saving config: {e}[/red]")
+
+    def update(self, **kwargs):
+        """Update config values."""
+        self.config.update(kwargs)
+        self.save()
+
+    @staticmethod
+    def update_proxy_api_key(new_key: str):
+        """Update PROXY_API_KEY in .env only"""
+        env_file = _get_env_file()
+        set_key(str(env_file), "PROXY_API_KEY", new_key)
+        load_dotenv(dotenv_path=env_file, override=True)
+
+
+class SettingsDetector:
+    """Detects settings from .env for display"""
+
+    @staticmethod
+    def _load_local_env() -> dict:
+        """Load environment variables from local .env file only"""
+        env_file = _get_env_file()
+        env_dict = {}
+        if not env_file.exists():
+            return env_dict
+        try:
+            with open(env_file, "r", encoding="utf-8") as f:
+                for line in f:
+                    line = line.strip()
+                    if not line or line.startswith("#"):
+                        continue
+                    if "=" in line:
+                        key, _, value = line.partition("=")
+                        key, value = key.strip(), value.strip()
+                        if value and value[0] in ('"', "'") and value[-1] == value[0]:
+                            value = value[1:-1]
+                        env_dict[key] = value
+        except (IOError, OSError):
+            pass
+        return env_dict
+
+    @staticmethod
+    def get_all_settings() -> dict:
+        """Returns comprehensive settings overview (includes provider_settings which triggers heavy imports)"""
+        return {
+            "credentials": SettingsDetector.detect_credentials(),
+            "custom_bases": SettingsDetector.detect_custom_api_bases(),
+            "model_definitions": SettingsDetector.detect_model_definitions(),
+            "concurrency_limits": SettingsDetector.detect_concurrency_limits(),
+            "model_filters": SettingsDetector.detect_model_filters(),
+            "provider_settings": SettingsDetector.detect_provider_settings(),
+        }
+
+    @staticmethod
+    def get_basic_settings() -> dict:
+        """Returns basic settings overview without provider_settings (avoids heavy imports)"""
+        return {
+            "credentials": SettingsDetector.detect_credentials(),
+            "custom_bases": SettingsDetector.detect_custom_api_bases(),
+            "model_definitions": SettingsDetector.detect_model_definitions(),
+            "concurrency_limits": SettingsDetector.detect_concurrency_limits(),
+            "model_filters": SettingsDetector.detect_model_filters(),
+        }
+
+    @staticmethod
+    def detect_credentials() -> dict:
+        """Detect API keys and OAuth credentials"""
+        import re
+        from pathlib import Path
+
+        providers = {}
+
+        # Scan for API keys
+        env_vars = SettingsDetector._load_local_env()
+        for key, value in env_vars.items():
+            if "_API_KEY" in key and key != "PROXY_API_KEY":
+                provider = key.split("_API_KEY")[0].lower()
+                if provider not in providers:
+                    providers[provider] = {"api_keys": 0, "oauth": 0, "custom": False}
+                providers[provider]["api_keys"] += 1
+
+        # Scan for file-based OAuth credentials
+        oauth_dir = Path("oauth_creds")
+        if oauth_dir.exists():
+            for file in oauth_dir.glob("*_oauth_*.json"):
+                provider = file.name.split("_oauth_")[0]
+                if provider not in providers:
+                    providers[provider] = {"api_keys": 0, "oauth": 0, "custom": False}
+                providers[provider]["oauth"] += 1
+
+        # Scan for env-based OAuth credentials
+        # Maps provider name to the ENV_PREFIX used by the provider
+        # (duplicated from credential_manager to avoid heavy imports)
+        env_oauth_providers = {
+            "gemini_cli": "GEMINI_CLI",
+            "antigravity": "ANTIGRAVITY",
+            "qwen_code": "QWEN_CODE",
+            "iflow": "IFLOW",
+        }
+
+        for provider, env_prefix in env_oauth_providers.items():
+            oauth_count = 0
+
+            # Check numbered credentials (PROVIDER_N_ACCESS_TOKEN pattern)
+            numbered_pattern = re.compile(rf"^{env_prefix}_(\d+)_ACCESS_TOKEN$")
+            for key in env_vars.keys():
+                match = numbered_pattern.match(key)
+                if match:
+                    index = match.group(1)
+                    refresh_key = f"{env_prefix}_{index}_REFRESH_TOKEN"
+                    if refresh_key in env_vars and env_vars[refresh_key]:
+                        oauth_count += 1
+
+            # Check legacy single credential (if no numbered found)
+            if oauth_count == 0:
+                access_key = f"{env_prefix}_ACCESS_TOKEN"
+                refresh_key = f"{env_prefix}_REFRESH_TOKEN"
+                if env_vars.get(access_key) and env_vars.get(refresh_key):
+                    oauth_count = 1
+
+            if oauth_count > 0:
+                if provider not in providers:
+                    providers[provider] = {"api_keys": 0, "oauth": 0, "custom": False}
+                providers[provider]["oauth"] += oauth_count
+
+        # Mark custom providers (have API_BASE set)
+        for provider in providers:
+            if os.getenv(f"{provider.upper()}_API_BASE"):
+                providers[provider]["custom"] = True
+
+        return providers
+
+    @staticmethod
+    def detect_custom_api_bases() -> dict:
+        """Detect custom API base URLs (not in hardcoded map)"""
+        from proxy_app.provider_urls import PROVIDER_URL_MAP
+
+        bases = {}
+        env_vars = SettingsDetector._load_local_env()
+        for key, value in env_vars.items():
+            if key.endswith("_API_BASE"):
+                provider = key.replace("_API_BASE", "").lower()
+                # Only include if NOT in hardcoded map
+                if provider not in PROVIDER_URL_MAP:
+                    bases[provider] = value
+        return bases
+
+    @staticmethod
+    def detect_model_definitions() -> dict:
+        """Detect provider model definitions"""
+        models = {}
+        env_vars = SettingsDetector._load_local_env()
+        for key, value in env_vars.items():
+            if key.endswith("_MODELS"):
+                provider = key.replace("_MODELS", "").lower()
+                try:
+                    parsed = json.loads(value)
+                    if isinstance(parsed, dict):
+                        models[provider] = len(parsed)
+                    elif isinstance(parsed, list):
+                        models[provider] = len(parsed)
+                except (json.JSONDecodeError, ValueError):
+                    pass
+        return models
+
+    @staticmethod
+    def detect_concurrency_limits() -> dict:
+        """Detect max concurrent requests per key"""
+        limits = {}
+        env_vars = SettingsDetector._load_local_env()
+        for key, value in env_vars.items():
+            if key.startswith("MAX_CONCURRENT_REQUESTS_PER_KEY_"):
+                provider = key.replace("MAX_CONCURRENT_REQUESTS_PER_KEY_", "").lower()
+                try:
+                    limits[provider] = int(value)
+                except (json.JSONDecodeError, ValueError):
+                    pass
+        return limits
+
+    @staticmethod
+    def detect_model_filters() -> dict:
+        """Detect active model filters (basic info only: defined or not)"""
+        filters = {}
+        env_vars = SettingsDetector._load_local_env()
+        for key, value in env_vars.items():
+            if key.startswith("IGNORE_MODELS_") or key.startswith("WHITELIST_MODELS_"):
+                filter_type = "ignore" if key.startswith("IGNORE") else "whitelist"
+                provider = key.replace(f"{filter_type.upper()}_MODELS_", "").lower()
+                if provider not in filters:
+                    filters[provider] = {"has_ignore": False, "has_whitelist": False}
+                if filter_type == "ignore":
+                    filters[provider]["has_ignore"] = True
+                else:
+                    filters[provider]["has_whitelist"] = True
+        return filters
+
+    @staticmethod
+    def detect_provider_settings() -> dict:
+        """Detect provider-specific settings (Antigravity, Gemini CLI)"""
+        try:
+            from proxy_app.settings_tool import PROVIDER_SETTINGS_MAP
+        except ImportError:
+            # Fallback for direct execution or testing
+            from .settings_tool import PROVIDER_SETTINGS_MAP
+
+        provider_settings = {}
+        env_vars = SettingsDetector._load_local_env()
+
+        for provider, definitions in PROVIDER_SETTINGS_MAP.items():
+            modified_count = 0
+            for key, definition in definitions.items():
+                env_value = env_vars.get(key)
+                if env_value is not None:
+                    # Check if value differs from default
+                    default = definition.get("default")
+                    setting_type = definition.get("type", "str")
+
+                    try:
+                        if setting_type == "bool":
+                            current = env_value.lower() in ("true", "1", "yes")
+                        elif setting_type == "int":
+                            current = int(env_value)
+                        else:
+                            current = env_value
+
+                        if current != default:
+                            modified_count += 1
+                    except (ValueError, AttributeError):
+                        pass
+
+            if modified_count > 0:
+                provider_settings[provider] = modified_count
+
+        return provider_settings
+
+
+class LauncherTUI:
+    """Main launcher interface"""
+
+    def __init__(self):
+        self.console = Console()
+        self.config = LauncherConfig()
+        self.running = True
+        self.env_file = _get_env_file()
+        # Load .env file to ensure environment variables are available
+        load_dotenv(dotenv_path=self.env_file, override=True)
+
+    def needs_onboarding(self) -> bool:
+        """Check if onboarding is needed"""
+        return not self.env_file.exists() or not os.getenv("PROXY_API_KEY")
+
+    def run(self):
+        """Main TUI loop"""
+        while self.running:
+            self.show_main_menu()
+
+    def show_main_menu(self):
+        """Display main menu and handle selection"""
+        clear_screen()
+
+        # Detect basic settings (excludes provider_settings to avoid heavy imports)
+        settings = SettingsDetector.get_basic_settings()
+        credentials = settings["credentials"]
+        custom_bases = settings["custom_bases"]
+
+        # Check if setup is needed
+        show_warning = self.needs_onboarding()
+
+        # Build title with GitHub link
+        self.console.print(
+            Panel.fit(
+                "[bold cyan]🚀 LLM API Key Proxy - Interactive Launcher[/bold cyan]",
+                border_style="cyan",
+            )
+        )
+        self.console.print(
+            "[dim]GitHub: [blue underline]https://github.com/Mirrowel/LLM-API-Key-Proxy[/blue underline][/dim]"
+        )
+
+        # Show warning if .env file doesn't exist
+        if show_warning:
+            self.console.print()
+            self.console.print(
+                Panel(
+                    Text.from_markup(
+                        "⚠️  [bold yellow]INITIAL SETUP REQUIRED[/bold yellow]\n\n"
+                        "The proxy needs initial configuration:\n"
+                        "  ❌ No .env file found\n\n"
+                        "Why this matters:\n"
+                        "  • The .env file stores your credentials and settings\n"
+                        "  • PROXY_API_KEY protects your proxy from unauthorized access\n"
+                        "  • Provider API keys enable LLM access\n\n"
+                        "What to do:\n"
+                        '  1. Select option "3. Manage Credentials" to launch the credential tool\n'
+                        "  2. The tool will create .env and set up PROXY_API_KEY automatically\n"
+                        "  3. You can add provider credentials (API keys or OAuth)\n\n"
+                        "⚠️  Note: The credential tool adds PROXY_API_KEY by default.\n"
+                        "   You can remove it later if you want an unsecured proxy."
+                    ),
+                    border_style="yellow",
+                    expand=False,
+                )
+            )
+        # Show security warning if PROXY_API_KEY is missing (but .env exists)
+        elif not os.getenv("PROXY_API_KEY"):
+            self.console.print()
+            self.console.print(
+                Panel(
+                    Text.from_markup(
+                        "⚠️  [bold red]SECURITY WARNING: PROXY_API_KEY Not Set[/bold red]\n\n"
+                        "Your proxy is currently UNSECURED!\n"
+                        "Anyone can access it without authentication.\n\n"
+                        "This is a serious security risk if your proxy is accessible\n"
+                        "from the internet or untrusted networks.\n\n"
+                        "👉 [bold]Recommended:[/bold] Set PROXY_API_KEY in .env file\n"
+                        '   Use option "2. Configure Proxy Settings" → "3. Set Proxy API Key"\n'
+                        '   or option "3. Manage Credentials"'
+                    ),
+                    border_style="red",
+                    expand=False,
+                )
+            )
+
+        # Show config
+        self.console.print()
+        self.console.print("[bold]📋 Proxy Configuration[/bold]")
+        self.console.print("━" * 70)
+        self.console.print(f"   Host:                {self.config.config['host']}")
+        self.console.print(f"   Port:                {self.config.config['port']}")
+        self.console.print(
+            f"   Transaction Logging: {'✅ Enabled' if self.config.config['enable_request_logging'] else '❌ Disabled'}"
+        )
+        self.console.print(
+            f"   Raw I/O Logging:     {'✅ Enabled' if self.config.config.get('enable_raw_logging', False) else '❌ Disabled'}"
+        )
+
+        # Show actual API key value
+        proxy_key = os.getenv("PROXY_API_KEY")
+        if proxy_key:
+            self.console.print(f"   Proxy API Key:       {proxy_key}")
+        else:
+            self.console.print("   Proxy API Key:       [red]Not Set (INSECURE!)[/red]")
+
+        # Show status summary
+        self.console.print()
+        self.console.print("[bold]📊 Status Summary[/bold]")
+        self.console.print("━" * 70)
+        provider_count = len(credentials)
+        custom_count = len(custom_bases)
+
+        self.console.print(f"   Providers:           {provider_count} configured")
+        self.console.print(f"   Custom Providers:    {custom_count} configured")
+        # Note: provider_settings detection is deferred to avoid heavy imports on startup
+        has_advanced = bool(
+            settings["model_definitions"]
+            or settings["concurrency_limits"]
+            or settings["model_filters"]
+        )
+        self.console.print(
+            f"   Advanced Settings:   {'Active (view in menu 4)' if has_advanced else 'None (view menu 4 for details)'}"
+        )
+
+        # Show menu
+        self.console.print()
+        self.console.print("━" * 70)
+        self.console.print()
+        self.console.print("[bold]🎯 Main Menu[/bold]")
+        self.console.print()
+        if show_warning:
+            self.console.print("   1. ▶️  Run Proxy Server")
+            self.console.print("   2. ⚙️  Configure Proxy Settings")
+            self.console.print(
+                "   3. 🔑 Manage Credentials            ⬅️  [bold yellow]Start here![/bold yellow]"
+            )
+        else:
+            self.console.print("   1. ▶️  Run Proxy Server")
+            self.console.print("   2. ⚙️  Configure Proxy Settings")
+            self.console.print("   3. 🔑 Manage Credentials")
+
+        self.console.print("   4. 📊 View Provider & Advanced Settings")
+        self.console.print("   5. 📈 View Quota & Usage Stats (Alpha)")
+        self.console.print("   6. 🔄 Reload Configuration")
+        self.console.print("   7. ℹ️  About")
+        self.console.print("   8. 🚪 Exit")
+
+        self.console.print()
+        self.console.print("━" * 70)
+        self.console.print()
+
+        choice = Prompt.ask(
+            "Select option",
+            choices=["1", "2", "3", "4", "5", "6", "7", "8"],
+            show_choices=False,
+        )
+
+        if choice == "1":
+            self.run_proxy()
+        elif choice == "2":
+            self.show_config_menu()
+        elif choice == "3":
+            self.launch_credential_tool()
+        elif choice == "4":
+            self.show_provider_settings_menu()
+        elif choice == "5":
+            self.launch_quota_viewer()
+        elif choice == "6":
+            load_dotenv(dotenv_path=_get_env_file(), override=True)
+            self.config = LauncherConfig()  # Reload config
+            self.console.print("\n[green]✅ Configuration reloaded![/green]")
+        elif choice == "7":
+            self.show_about()
+        elif choice == "8":
+            self.running = False
+            sys.exit(0)
+
+    def confirm_setting_change(self, setting_name: str, warning_lines: list) -> bool:
+        """
+        Display a warning and require Y/N (case-sensitive) confirmation.
+        Re-prompts until user enters exactly 'Y' or 'N'.
+        Returns True only if user enters 'Y'.
+        """
+        clear_screen()
+        self.console.print()
+        self.console.print(
+            Panel(
+                Text.from_markup(
+                    f"[bold yellow]⚠️  WARNING: You are about to change the {setting_name}[/bold yellow]\n\n"
+                    + "\n".join(warning_lines)
+                    + "\n\n[bold]If you are not sure about changing this - don't.[/bold]"
+                ),
+                border_style="yellow",
+                expand=False,
+            )
+        )
+
+        while True:
+            response = Prompt.ask(
+                "Enter [bold]Y[/bold] to confirm, [bold]N[/bold] to cancel (case-sensitive)"
+            )
+            if response == "Y":
+                return True
+            elif response == "N":
+                self.console.print("\n[dim]Operation cancelled.[/dim]")
+                return False
+            else:
+                self.console.print(
+                    "[red]Please enter exactly 'Y' or 'N' (case-sensitive)[/red]"
+                )
+
+    def show_config_menu(self):
+        """Display configuration sub-menu"""
+        while True:
+            clear_screen()
+
+            self.console.print(
+                Panel.fit(
+                    "[bold cyan]⚙️  Proxy Configuration[/bold cyan]", border_style="cyan"
+                )
+            )
+
+            self.console.print()
+            self.console.print("[bold]📋 Current Settings[/bold]")
+            self.console.print("━" * 70)
+            self.console.print(f"   Host:                {self.config.config['host']}")
+            self.console.print(f"   Port:                {self.config.config['port']}")
+            self.console.print(
+                f"   Transaction Logging: {'✅ Enabled' if self.config.config['enable_request_logging'] else '❌ Disabled'}"
+            )
+            self.console.print(
+                f"   Raw I/O Logging:     {'✅ Enabled' if self.config.config.get('enable_raw_logging', False) else '❌ Disabled'}"
+            )
+            self.console.print(
+                f"   Proxy API Key:       {'✅ Set' if os.getenv('PROXY_API_KEY') else '❌ Not Set'}"
+            )
+
+            self.console.print()
+            self.console.print("━" * 70)
+            self.console.print()
+            self.console.print("[bold]⚙️  Configuration Options[/bold]")
+            self.console.print()
+            self.console.print("   1. 🌐 Set Host IP")
+            self.console.print("   2. 🔌 Set Port")
+            self.console.print("   3. 🔑 Set Proxy API Key")
+            self.console.print("   4. 📝 Toggle Transaction Logging")
+            self.console.print("   5. 📋 Toggle Raw I/O Logging")
+            self.console.print("   6. 🔄 Reset to Default Settings")
+            self.console.print("   7. ↩️  Back to Main Menu")
+
+            self.console.print()
+            self.console.print("━" * 70)
+            self.console.print()
+
+            choice = Prompt.ask(
+                "Select option",
+                choices=["1", "2", "3", "4", "5", "6", "7"],
+                show_choices=False,
+            )
+
+            if choice == "1":
+                # Show warning and require confirmation
+                confirmed = self.confirm_setting_change(
+                    "Host IP",
+                    [
+                        "Changing the host IP affects which network interfaces the proxy listens on:",
+                        "  • [cyan]127.0.0.1[/cyan] = Local access only (recommended for development)",
+                        "  • [cyan]0.0.0.0[/cyan] = Accessible from all network interfaces",
+                        "",
+                        "Applications configured to connect to the old host may fail to connect.",
+                    ],
+                )
+                if not confirmed:
+                    continue
+
+                new_host = Prompt.ask(
+                    "Enter new host IP", default=self.config.config["host"]
+                )
+                self.config.update(host=new_host)
+                self.console.print(f"\n[green]✅ Host updated to: {new_host}[/green]")
+            elif choice == "2":
+                # Show warning and require confirmation
+                confirmed = self.confirm_setting_change(
+                    "Port",
+                    [
+                        "Changing the port will affect all applications currently configured",
+                        "to connect to your proxy on the existing port.",
+                        "",
+                        "Applications using the old port will fail to connect.",
+                    ],
+                )
+                if not confirmed:
+                    continue
+
+                new_port = IntPrompt.ask(
+                    "Enter new port", default=self.config.config["port"]
+                )
+                if 1 <= new_port <= 65535:
+                    self.config.update(port=new_port)
+                    self.console.print(
+                        f"\n[green]✅ Port updated to: {new_port}[/green]"
+                    )
+                else:
+                    self.console.print("\n[red]❌ Port must be between 1-65535[/red]")
+            elif choice == "3":
+                # Show warning and require confirmation
+                confirmed = self.confirm_setting_change(
+                    "Proxy API Key",
+                    [
+                        "This is the authentication key that applications use to access your proxy.",
+                        "",
+                        "[bold red]⚠️  Changing this will BREAK all applications currently configured",
+                        "   with the existing API key![/bold red]",
+                        "",
+                        "[bold cyan]💡 If you want to add provider API keys (OpenAI, Gemini, etc.),",
+                        '   go to "3. 🔑 Manage Credentials" in the main menu instead.[/bold cyan]',
+                    ],
+                )
+                if not confirmed:
+                    continue
+
+                current = os.getenv("PROXY_API_KEY", "")
+                new_key = Prompt.ask(
+                    "Enter new Proxy API Key (leave empty to disable authentication)",
+                    default=current,
+                )
+
+                if new_key != current:
+                    # If setting to empty, show additional warning
+                    if not new_key:
+                        self.console.print(
+                            "\n[bold red]⚠️  Authentication will be DISABLED - anyone can access your proxy![/bold red]"
+                        )
+                        Prompt.ask("Press Enter to continue", default="")
+
+                    LauncherConfig.update_proxy_api_key(new_key)
+
+                    if new_key:
+                        self.console.print(
+                            "\n[green]✅ Proxy API Key updated successfully![/green]"
+                        )
+                        self.console.print("   Updated in .env file")
+                    else:
+                        self.console.print(
+                            "\n[yellow]⚠️  Proxy API Key cleared - authentication disabled![/yellow]"
+                        )
+                        self.console.print("   Updated in .env file")
+                else:
+                    self.console.print("\n[yellow]No changes made[/yellow]")
+            elif choice == "4":
+                current = self.config.config["enable_request_logging"]
+                self.config.update(enable_request_logging=not current)
+                self.console.print(
+                    f"\n[green]✅ Transaction Logging {'enabled' if not current else 'disabled'}![/green]"
+                )
+            elif choice == "5":
+                current = self.config.config.get("enable_raw_logging", False)
+                self.config.update(enable_raw_logging=not current)
+                self.console.print(
+                    f"\n[green]✅ Raw I/O Logging {'enabled' if not current else 'disabled'}![/green]"
+                )
+            elif choice == "6":
+                # Reset to Default Settings
+                # Define defaults
+                default_host = "127.0.0.1"
+                default_port = 8000
+                default_logging = False
+                default_raw_logging = False
+                default_api_key = "VerysecretKey"
+
+                # Get current values
+                current_host = self.config.config["host"]
+                current_port = self.config.config["port"]
+                current_logging = self.config.config["enable_request_logging"]
+                current_raw_logging = self.config.config.get(
+                    "enable_raw_logging", False
+                )
+                current_api_key = os.getenv("PROXY_API_KEY", "")
+
+                # Build comparison table
+                warning_lines = [
+                    "This will reset ALL proxy settings to their defaults:",
+                    "",
+                    "[bold]   Setting              Current Value         →  Default Value[/bold]",
+                    "   " + "─" * 62,
+                    f"   Host IP              {current_host:20} →  {default_host}",
+                    f"   Port                 {str(current_port):20} →  {default_port}",
+                    f"   Transaction Logging  {'Enabled':20} →  Disabled"
+                    if current_logging
+                    else f"   Transaction Logging  {'Disabled':20} →  Disabled",
+                    f"   Raw I/O Logging      {'Enabled':20} →  Disabled"
+                    if current_raw_logging
+                    else f"   Raw I/O Logging      {'Disabled':20} →  Disabled",
+                    f"   Proxy API Key        {current_api_key[:20]:20} →  {default_api_key}",
+                    "",
+                    "[bold red]⚠️  This may break applications configured with current settings![/bold red]",
+                ]
+
+                confirmed = self.confirm_setting_change(
+                    "Settings (Reset to Defaults)", warning_lines
+                )
+                if not confirmed:
+                    continue
+
+                # Apply defaults
+                self.config.update(
+                    host=default_host,
+                    port=default_port,
+                    enable_request_logging=default_logging,
+                    enable_raw_logging=default_raw_logging,
+                )
+                LauncherConfig.update_proxy_api_key(default_api_key)
+
+                self.console.print(
+                    "\n[green]✅ All settings have been reset to defaults![/green]"
+                )
+                self.console.print(f"   Host:               {default_host}")
+                self.console.print(f"   Port:               {default_port}")
+                self.console.print(f"   Transaction Logging: Disabled")
+                self.console.print(f"   Raw I/O Logging:    Disabled")
+                self.console.print(f"   Proxy API Key:      {default_api_key}")
+            elif choice == "7":
+                break
+
+    def show_provider_settings_menu(self):
+        """Display provider/advanced settings (read-only + launch tool)"""
+        clear_screen()
+
+        # Use basic settings to avoid heavy imports - provider_settings deferred to Settings Tool
+        settings = SettingsDetector.get_basic_settings()
+
+        credentials = settings["credentials"]
+        custom_bases = settings["custom_bases"]
+        model_defs = settings["model_definitions"]
+        concurrency = settings["concurrency_limits"]
+        filters = settings["model_filters"]
+
+        self.console.print(
+            Panel.fit(
+                "[bold cyan]📊 Provider & Advanced Settings[/bold cyan]",
+                border_style="cyan",
+            )
+        )
+
+        # Configured Providers
+        self.console.print()
+        self.console.print("[bold]📊 Configured Providers[/bold]")
+        self.console.print("━" * 70)
+        if credentials:
+            for provider, info in credentials.items():
+                provider_name = provider.title()
+                parts = []
+                if info["api_keys"] > 0:
+                    parts.append(
+                        f"{info['api_keys']} API key{'s' if info['api_keys'] > 1 else ''}"
+                    )
+                if info["oauth"] > 0:
+                    parts.append(
+                        f"{info['oauth']} OAuth credential{'s' if info['oauth'] > 1 else ''}"
+                    )
+
+                display = " + ".join(parts)
+                if info["custom"]:
+                    display += " (Custom)"
+
+                self.console.print(f"   ✅ {provider_name:20} {display}")
+        else:
+            self.console.print("   [dim]No providers configured[/dim]")
+
+        # Custom API Bases
+        if custom_bases:
+            self.console.print()
+            self.console.print("[bold]🌐 Custom API Bases[/bold]")
+            self.console.print("━" * 70)
+            for provider, base in custom_bases.items():
+                self.console.print(f"   • {provider:15} {base}")
+
+        # Model Definitions
+        if model_defs:
+            self.console.print()
+            self.console.print("[bold]📦 Provider Model Definitions[/bold]")
+            self.console.print("━" * 70)
+            for provider, count in model_defs.items():
+                self.console.print(
+                    f"   • {provider:15} {count} model{'s' if count > 1 else ''} configured"
+                )
+
+        # Concurrency Limits
+        if concurrency:
+            self.console.print()
+            self.console.print("[bold]⚡ Concurrency Limits[/bold]")
+            self.console.print("━" * 70)
+            for provider, limit in concurrency.items():
+                self.console.print(f"   • {provider:15} {limit} requests/key")
+            self.console.print("   • Default:        1 request/key (all others)")
+
+        # Model Filters (basic info only)
+        if filters:
+            self.console.print()
+            self.console.print("[bold]🎯 Model Filters[/bold]")
+            self.console.print("━" * 70)
+            for provider, filter_info in filters.items():
+                status_parts = []
+                if filter_info["has_whitelist"]:
+                    status_parts.append("Whitelist")
+                if filter_info["has_ignore"]:
+                    status_parts.append("Ignore list")
+                status = " + ".join(status_parts) if status_parts else "None"
+                self.console.print(f"   • {provider:15} ✅ {status}")
+
+        # Provider-Specific Settings (deferred to Settings Tool to avoid heavy imports)
+        self.console.print()
+        self.console.print("[bold]🔬 Provider-Specific Settings[/bold]")
+        self.console.print("━" * 70)
+        self.console.print(
+            "   [dim]Launch Settings Tool to view/configure provider-specific settings[/dim]"
+        )
+
+        # Actions
+        self.console.print()
+        self.console.print("━" * 70)
+        self.console.print()
+        self.console.print("[bold]💡 Actions[/bold]")
+        self.console.print()
+        self.console.print(
+            "   1. 🔧 Launch Settings Tool      (configure advanced settings)"
+        )
+        self.console.print("   2. ↩️  Back to Main Menu")
+
+        self.console.print()
+        self.console.print("━" * 70)
+        self.console.print(
+            "[dim]ℹ️  Advanced settings are stored in .env file.\n   Use the Settings Tool to configure them interactively.[/dim]"
+        )
+        self.console.print()
+        self.console.print(
+            "[dim]⚠️  Note: Settings Tool supports only common configuration types.\n   For complex settings, edit .env directly.[/dim]"
+        )
+        self.console.print()
+
+        choice = Prompt.ask("Select option", choices=["1", "2"], show_choices=False)
+
+        if choice == "1":
+            self.launch_settings_tool()
+        # choice == "2" returns to main menu
+
+    def launch_credential_tool(self):
+        """Launch credential management tool"""
+        import time
+
+        # CRITICAL: Show full loading UI to replace the 6-7 second blank wait
+        clear_screen()
+
+        _start_time = time.time()
+
+        # Show the same header as standalone mode
+        self.console.print("━" * 70)
+        self.console.print("Interactive Credential Setup Tool")
+        self.console.print("GitHub: https://github.com/Mirrowel/LLM-API-Key-Proxy")
+        self.console.print("━" * 70)
+        self.console.print("Loading credential management components...")
+
+        # Now import with spinner (this is where the 6-7 second delay happens)
+        with self.console.status("Initializing credential tool...", spinner="dots"):
+            from rotator_library.credential_tool import (
+                run_credential_tool,
+                _ensure_providers_loaded,
+            )
+
+            _, PROVIDER_PLUGINS = _ensure_providers_loaded()
+        self.console.print("✓ Credential tool initialized")
+
+        _elapsed = time.time() - _start_time
+        self.console.print(
+            f"✓ Tool ready in {_elapsed:.2f}s ({len(PROVIDER_PLUGINS)} providers available)"
+        )
+
+        # Small delay to let user see the ready message
+        time.sleep(0.5)
+
+        # Run the tool with from_launcher=True to skip duplicate loading screen
+        run_credential_tool(from_launcher=True)
+        # Reload environment after credential tool
+        load_dotenv(dotenv_path=_get_env_file(), override=True)
+
+    def launch_settings_tool(self):
+        """Launch settings configuration tool"""
+        import time
+
+        clear_screen()
+
+        self.console.print("━" * 70)
+        self.console.print("Advanced Settings Configuration Tool")
+        self.console.print("━" * 70)
+
+        _start_time = time.time()
+
+        with self.console.status("Initializing settings tool...", spinner="dots"):
+            from proxy_app.settings_tool import run_settings_tool
+
+        _elapsed = time.time() - _start_time
+        self.console.print(f"✓ Settings tool ready in {_elapsed:.2f}s")
+
+        time.sleep(0.3)
+
+        run_settings_tool()
+        # Reload environment after settings tool
+        load_dotenv(dotenv_path=_get_env_file(), override=True)
+
+    def launch_quota_viewer(self):
+        """Launch the quota stats viewer"""
+        clear_screen()
+
+        self.console.print("━" * 70)
+        self.console.print("Quota & Usage Statistics Viewer")
+        self.console.print("━" * 70)
+        self.console.print()
+
+        # Import the lightweight viewer (no heavy imports)
+        from proxy_app.quota_viewer import run_quota_viewer
+
+        run_quota_viewer()
+
+    def show_about(self):
+        """Display About page with project information"""
+        clear_screen()
+
+        self.console.print(
+            Panel.fit(
+                "[bold cyan]ℹ️  About LLM API Key Proxy[/bold cyan]", border_style="cyan"
+            )
+        )
+
+        self.console.print()
+        self.console.print("[bold]📦 Project Information[/bold]")
+        self.console.print("━" * 70)
+        self.console.print("   [bold cyan]LLM API Key Proxy[/bold cyan]")
+        self.console.print(
+            "   A lightweight, high-performance proxy server for managing"
+        )
+        self.console.print("   LLM API keys with automatic rotation and OAuth support")
+        self.console.print()
+        self.console.print(
+            "   [dim]GitHub:[/dim] [blue underline]https://github.com/Mirrowel/LLM-API-Key-Proxy[/blue underline]"
+        )
+
+        self.console.print()
+        self.console.print("[bold]✨ Key Features[/bold]")
+        self.console.print("━" * 70)
+        self.console.print(
+            "   • [green]Smart Key Rotation[/green] - Automatic rotation across multiple API keys"
+        )
+        self.console.print(
+            "   • [green]OAuth Support[/green] - Automated OAuth flows for supported providers"
+        )
+        self.console.print(
+            "   • [green]Multiple Providers[/green] - Support for 10+ LLM providers"
+        )
+        self.console.print(
+            "   • [green]Custom Providers[/green] - Easy integration of custom OpenAI-compatible APIs"
+        )
+        self.console.print(
+            "   • [green]Advanced Filtering[/green] - Model whitelists and ignore lists per provider"
+        )
+        self.console.print(
+            "   • [green]Concurrency Control[/green] - Per-key rate limiting and request management"
+        )
+        self.console.print(
+            "   • [green]Cost Tracking[/green] - Track usage and costs across all providers"
+        )
+        self.console.print(
+            "   • [green]Interactive TUI[/green] - Beautiful terminal interface for easy configuration"
+        )
+
+        self.console.print()
+        self.console.print("[bold]📝 License & Credits[/bold]")
+        self.console.print("━" * 70)
+        self.console.print("   Made with ❤️  by the community")
+        self.console.print("   Open source - contributions welcome!")
+
+        self.console.print()
+        self.console.print("━" * 70)
+        self.console.print()
+
+        Prompt.ask("Press Enter to return to main menu", default="")
+
+    def run_proxy(self):
+        """Prepare and launch proxy in same window"""
+        # Check if forced onboarding needed
+        if self.needs_onboarding():
+            clear_screen()
+            self.console.print(
+                Panel(
+                    Text.from_markup(
+                        "⚠️  [bold yellow]Setup Required[/bold yellow]\n\n"
+                        "Cannot start without .env.\n"
+                        "Launching credential tool..."
+                    ),
+                    border_style="yellow",
+                )
+            )
+
+            # Force credential tool
+            from rotator_library.credential_tool import (
+                ensure_env_defaults,
+                run_credential_tool,
+            )
+
+            ensure_env_defaults()
+            load_dotenv(dotenv_path=_get_env_file(), override=True)
+            run_credential_tool()
+            load_dotenv(dotenv_path=_get_env_file(), override=True)
+
+            # Check again after credential tool
+            if not os.getenv("PROXY_API_KEY"):
+                self.console.print(
+                    "\n[red]❌ PROXY_API_KEY still not set. Cannot start proxy.[/red]"
+                )
+                return
+
+        # Clear console and modify sys.argv
+        clear_screen()
+        self.console.print(
+            f"\n[bold green]🚀 Starting proxy on {self.config.config['host']}:{self.config.config['port']}...[/bold green]\n"
+        )
+
+        # Brief pause so user sees the message before main.py takes over
+        import time
+
+        time.sleep(0.5)
+
+        # Reconstruct sys.argv for main.py
+        sys.argv = [
+            "main.py",
+            "--host",
+            self.config.config["host"],
+            "--port",
+            str(self.config.config["port"]),
+        ]
+        if self.config.config["enable_request_logging"]:
+            sys.argv.append("--enable-request-logging")
+        if self.config.config.get("enable_raw_logging", False):
+            sys.argv.append("--enable-raw-logging")
+
+        # Exit TUI - main.py will continue execution
+        self.running = False
+
+
+def run_launcher_tui():
+    """Entry point for launcher TUI"""
+    tui = LauncherTUI()
+    tui.run()

src/proxy_app/main.py

@@ -0,0 +1,1731 @@
+# SPDX-License-Identifier: MIT
+# Copyright (c) 2026 Mirrowel
+
+import time
+import uuid
+
+# Phase 1: Minimal imports for arg parsing and TUI
+import asyncio
+import os
+from pathlib import Path
+import sys
+import argparse
+import logging
+
+# --- Argument Parsing (BEFORE heavy imports) ---
+parser = argparse.ArgumentParser(description="API Key Proxy Server")
+parser.add_argument(
+    "--host", type=str, default="0.0.0.0", help="Host to bind the server to."
+)
+parser.add_argument("--port", type=int, default=8000, help="Port to run the server on.")
+parser.add_argument(
+    "--enable-request-logging",
+    action="store_true",
+    help="Enable transaction logging in the library (logs request/response with provider correlation).",
+)
+parser.add_argument(
+    "--enable-raw-logging",
+    action="store_true",
+    help="Enable raw I/O logging at proxy boundary (captures unmodified HTTP data, disabled by default).",
+)
+parser.add_argument(
+    "--add-credential",
+    action="store_true",
+    help="Launch the interactive tool to add a new OAuth credential.",
+)
+args, _ = parser.parse_known_args()
+
+# Add the 'src' directory to the Python path
+sys.path.append(str(Path(__file__).resolve().parent.parent))
+
+# Check if we should launch TUI (no arguments = TUI mode)
+if len(sys.argv) == 1:
+    # TUI MODE - Load ONLY what's needed for the launcher (fast path!)
+    from proxy_app.launcher_tui import run_launcher_tui
+
+    run_launcher_tui()
+    # Launcher modifies sys.argv and returns, or exits if user chose Exit
+    # If we get here, user chose "Run Proxy" and sys.argv is modified
+    # Re-parse arguments with modified sys.argv
+    args = parser.parse_args()
+
+# Check if credential tool mode (also doesn't need heavy proxy imports)
+if args.add_credential:
+    from rotator_library.credential_tool import run_credential_tool
+
+    run_credential_tool()
+    sys.exit(0)
+
+# If we get here, we're ACTUALLY running the proxy - NOW show startup messages and start timer
+_start_time = time.time()
+
+# Load all .env files from root folder (main .env first, then any additional *.env files)
+from dotenv import load_dotenv
+from glob import glob
+
+# Get the application root directory (EXE dir if frozen, else CWD)
+# Inlined here to avoid triggering heavy rotator_library imports before loading screen
+if getattr(sys, "frozen", False):
+    _root_dir = Path(sys.executable).parent
+else:
+    _root_dir = Path.cwd()
+
+# [HUGGING FACE SUPPORT] If a bulk environment block is provided via Secret, save it to a file
+# This allows users to paste their entire .env content into a single HF Secret called CONFIG_ENV
+_bulk_env = os.getenv("CONFIG_ENV")
+if _bulk_env:
+    _bulk_env_file = _root_dir / "bulk_config.env"
+    with open(_bulk_env_file, "w", encoding="utf-8") as _f:
+        _f.write(_bulk_env)
+    print(f"📝 Detected 'CONFIG_ENV' secret, saved to '{_bulk_env_file.name}'")
+
+# Load main .env first
+load_dotenv(_root_dir / ".env")
+
+# Load any additional .env files (e.g., antigravity_all_combined.env, gemini_cli_all_combined.env)
+_env_files_found = list(_root_dir.glob("*.env"))
+for _env_file in sorted(_root_dir.glob("*.env")):
+    if _env_file.name != ".env":  # Skip main .env (already loaded)
+        load_dotenv(_env_file, override=False)  # Don't override existing values
+
+# Log discovered .env files for deployment verification
+if _env_files_found:
+    _env_names = [_ef.name for _ef in _env_files_found]
+    print(f"📁 Loaded {len(_env_files_found)} .env file(s): {', '.join(_env_names)}")
+
+# Get proxy API key for display
+proxy_api_key = os.getenv("PROXY_API_KEY")
+if proxy_api_key:
+    key_display = f"✓ {proxy_api_key}"
+else:
+    key_display = "✗ Not Set (INSECURE - anyone can access!)"
+
+print("━" * 70)
+print(f"Starting proxy on {args.host}:{args.port}")
+print(f"Proxy API Key: {key_display}")
+print(f"GitHub: https://github.com/Mirrowel/LLM-API-Key-Proxy")
+print("━" * 70)
+print("Loading server components...")
+
+
+# Phase 2: Load Rich for loading spinner (lightweight)
+from rich.console import Console
+
+_console = Console()
+
+# Phase 3: Heavy dependencies with granular loading messages
+print("  → Loading FastAPI framework...")
+with _console.status("[dim]Loading FastAPI framework...", spinner="dots"):
+    from contextlib import asynccontextmanager
+    from fastapi import FastAPI, Request, HTTPException, Depends
+    from fastapi.middleware.cors import CORSMiddleware
+    from fastapi.responses import StreamingResponse, JSONResponse
+    from fastapi.security import APIKeyHeader
+
+print("  → Loading core dependencies...")
+with _console.status("[dim]Loading core dependencies...", spinner="dots"):
+    from dotenv import load_dotenv
+    import colorlog
+    import json
+    from typing import AsyncGenerator, Any, List, Optional, Union
+    from pydantic import BaseModel, ConfigDict, Field
+
+    # --- Early Log Level Configuration ---
+    logging.getLogger("LiteLLM").setLevel(logging.WARNING)
+
+print("  → Loading LiteLLM library...")
+with _console.status("[dim]Loading LiteLLM library...", spinner="dots"):
+    import litellm
+
+# Phase 4: Application imports with granular loading messages
+print("  → Initializing proxy core...")
+with _console.status("[dim]Initializing proxy core...", spinner="dots"):
+    from rotator_library import RotatingClient
+    from rotator_library.credential_manager import CredentialManager
+    from rotator_library.background_refresher import BackgroundRefresher
+    from rotator_library.model_info_service import init_model_info_service
+    from proxy_app.request_logger import log_request_to_console
+    from proxy_app.batch_manager import EmbeddingBatcher
+    from proxy_app.detailed_logger import RawIOLogger
+
+print("  → Discovering provider plugins...")
+# Provider lazy loading happens during import, so time it here
+_provider_start = time.time()
+with _console.status("[dim]Discovering provider plugins...", spinner="dots"):
+    from rotator_library import (
+        PROVIDER_PLUGINS,
+    )  # This triggers lazy load via __getattr__
+_provider_time = time.time() - _provider_start
+
+# Get count after import (without timing to avoid double-counting)
+_plugin_count = len(PROVIDER_PLUGINS)
+
+
+# --- Pydantic Models ---
+class EmbeddingRequest(BaseModel):
+    model: str
+    input: Union[str, List[str]]
+    input_type: Optional[str] = None
+    dimensions: Optional[int] = None
+    user: Optional[str] = None
+
+
+class ModelCard(BaseModel):
+    """Basic model card for minimal response."""
+
+    id: str
+    object: str = "model"
+    created: int = Field(default_factory=lambda: int(time.time()))
+    owned_by: str = "Mirro-Proxy"
+
+
+class ModelCapabilities(BaseModel):
+    """Model capability flags."""
+
+    tool_choice: bool = False
+    function_calling: bool = False
+    reasoning: bool = False
+    vision: bool = False
+    system_messages: bool = True
+    prompt_caching: bool = False
+    assistant_prefill: bool = False
+
+
+class EnrichedModelCard(BaseModel):
+    """Extended model card with pricing and capabilities."""
+
+    id: str
+    object: str = "model"
+    created: int = Field(default_factory=lambda: int(time.time()))
+    owned_by: str = "unknown"
+    # Pricing (optional - may not be available for all models)
+    input_cost_per_token: Optional[float] = None
+    output_cost_per_token: Optional[float] = None
+    cache_read_input_token_cost: Optional[float] = None
+    cache_creation_input_token_cost: Optional[float] = None
+    # Limits (optional)
+    max_input_tokens: Optional[int] = None
+    max_output_tokens: Optional[int] = None
+    context_window: Optional[int] = None
+    # Capabilities
+    mode: str = "chat"
+    supported_modalities: List[str] = Field(default_factory=lambda: ["text"])
+    supported_output_modalities: List[str] = Field(default_factory=lambda: ["text"])
+    capabilities: Optional[ModelCapabilities] = None
+    # Debug info (optional)
+    _sources: Optional[List[str]] = None
+    _match_type: Optional[str] = None
+
+    model_config = ConfigDict(extra="allow")  # Allow extra fields from the service
+
+
+class ModelList(BaseModel):
+    """List of models response."""
+
+    object: str = "list"
+    data: List[ModelCard]
+
+
+class EnrichedModelList(BaseModel):
+    """List of enriched models with pricing and capabilities."""
+
+    object: str = "list"
+    data: List[EnrichedModelCard]
+
+
+# --- Anthropic API Models (imported from library) ---
+from rotator_library.anthropic_compat import (
+    AnthropicMessagesRequest,
+    AnthropicCountTokensRequest,
+)
+
+
+# Calculate total loading time
+_elapsed = time.time() - _start_time
+print(
+    f"✓ Server ready in {_elapsed:.2f}s ({_plugin_count} providers discovered in {_provider_time:.2f}s)"
+)
+
+# Clear screen and reprint header for clean startup view
+# This pushes loading messages up (still in scroll history) but shows a clean final screen
+import os as _os_module
+
+_os_module.system("cls" if _os_module.name == "nt" else "clear")
+
+# Reprint header
+print("━" * 70)
+print(f"Starting proxy on {args.host}:{args.port}")
+print(f"Proxy API Key: {key_display}")
+print(f"GitHub: https://github.com/Mirrowel/LLM-API-Key-Proxy")
+print("━" * 70)
+print(
+    f"✓ Server ready in {_elapsed:.2f}s ({_plugin_count} providers discovered in {_provider_time:.2f}s)"
+)
+
+
+# Note: Debug logging will be added after logging configuration below
+
+# --- Logging Configuration ---
+# Import path utilities here (after loading screen) to avoid triggering heavy imports early
+from rotator_library.utils.paths import get_logs_dir, get_data_file
+
+LOG_DIR = get_logs_dir(_root_dir)
+
+# Configure a console handler with color (INFO and above only, no DEBUG)
+console_handler = colorlog.StreamHandler(sys.stdout)
+console_handler.setLevel(logging.INFO)
+formatter = colorlog.ColoredFormatter(
+    "%(log_color)s%(message)s",
+    log_colors={
+        "DEBUG": "cyan",
+        "INFO": "green",
+        "WARNING": "yellow",
+        "ERROR": "red",
+        "CRITICAL": "red,bg_white",
+    },
+)
+console_handler.setFormatter(formatter)
+
+# Configure a file handler for INFO-level logs and higher
+info_file_handler = logging.FileHandler(LOG_DIR / "proxy.log", encoding="utf-8")
+info_file_handler.setLevel(logging.INFO)
+info_file_handler.setFormatter(
+    logging.Formatter("%(asctime)s - %(name)s - %(levelname)s - %(message)s")
+)
+
+# Configure a dedicated file handler for all DEBUG-level logs
+debug_file_handler = logging.FileHandler(LOG_DIR / "proxy_debug.log", encoding="utf-8")
+debug_file_handler.setLevel(logging.DEBUG)
+debug_file_handler.setFormatter(
+    logging.Formatter("%(asctime)s - %(name)s - %(levelname)s - %(message)s")
+)
+
+
+# Create a filter to ensure the debug handler ONLY gets DEBUG messages from the rotator_library
+class RotatorDebugFilter(logging.Filter):
+    def filter(self, record):
+        return record.levelno == logging.DEBUG and record.name.startswith(
+            "rotator_library"
+        )
+
+
+debug_file_handler.addFilter(RotatorDebugFilter())
+
+# Configure a console handler with color
+console_handler = colorlog.StreamHandler(sys.stdout)
+console_handler.setLevel(logging.INFO)
+formatter = colorlog.ColoredFormatter(
+    "%(log_color)s%(message)s",
+    log_colors={
+        "DEBUG": "cyan",
+        "INFO": "green",
+        "WARNING": "yellow",
+        "ERROR": "red",
+        "CRITICAL": "red,bg_white",
+    },
+)
+console_handler.setFormatter(formatter)
+
+
+# Add a filter to prevent any LiteLLM logs from cluttering the console
+class NoLiteLLMLogFilter(logging.Filter):
+    def filter(self, record):
+        return not record.name.startswith("LiteLLM")
+
+
+console_handler.addFilter(NoLiteLLMLogFilter())
+
+# Get the root logger and set it to DEBUG to capture all messages
+root_logger = logging.getLogger()
+root_logger.setLevel(logging.DEBUG)
+
+# Add all handlers to the root logger
+root_logger.addHandler(info_file_handler)
+root_logger.addHandler(console_handler)
+root_logger.addHandler(debug_file_handler)
+
+# Silence other noisy loggers by setting their level higher than root
+logging.getLogger("uvicorn").setLevel(logging.WARNING)
+logging.getLogger("httpx").setLevel(logging.WARNING)
+
+# Isolate LiteLLM's logger to prevent it from reaching the console.
+# We will capture its logs via the logger_fn callback in the client instead.
+litellm_logger = logging.getLogger("LiteLLM")
+litellm_logger.handlers = []
+litellm_logger.propagate = False
+
+# Now that logging is configured, log the module load time to debug file only
+logging.debug(f"Modules loaded in {_elapsed:.2f}s")
+
+# Load environment variables from .env file
+load_dotenv(_root_dir / ".env")
+
+# --- Configuration ---
+USE_EMBEDDING_BATCHER = False
+ENABLE_REQUEST_LOGGING = args.enable_request_logging
+ENABLE_RAW_LOGGING = args.enable_raw_logging
+if ENABLE_REQUEST_LOGGING:
+    logging.info(
+        "Transaction logging is enabled (library-level with provider correlation)."
+    )
+if ENABLE_RAW_LOGGING:
+    logging.info("Raw I/O logging is enabled (proxy boundary, unmodified HTTP data).")
+PROXY_API_KEY = os.getenv("PROXY_API_KEY")
+# Note: PROXY_API_KEY validation moved to server startup to allow credential tool to run first
+
+# Discover API keys from environment variables
+api_keys = {}
+for key, value in os.environ.items():
+    if "_API_KEY" in key and key != "PROXY_API_KEY":
+        provider = key.split("_API_KEY")[0].lower()
+        if provider not in api_keys:
+            api_keys[provider] = []
+        api_keys[provider].append(value)
+
+# Load model ignore lists from environment variables
+ignore_models = {}
+for key, value in os.environ.items():
+    if key.startswith("IGNORE_MODELS_"):
+        provider = key.replace("IGNORE_MODELS_", "").lower()
+        models_to_ignore = [
+            model.strip() for model in value.split(",") if model.strip()
+        ]
+        ignore_models[provider] = models_to_ignore
+        logging.debug(
+            f"Loaded ignore list for provider '{provider}': {models_to_ignore}"
+        )
+
+# Load model whitelist from environment variables
+whitelist_models = {}
+for key, value in os.environ.items():
+    if key.startswith("WHITELIST_MODELS_"):
+        provider = key.replace("WHITELIST_MODELS_", "").lower()
+        models_to_whitelist = [
+            model.strip() for model in value.split(",") if model.strip()
+        ]
+        whitelist_models[provider] = models_to_whitelist
+        logging.debug(
+            f"Loaded whitelist for provider '{provider}': {models_to_whitelist}"
+        )
+
+# Load max concurrent requests per key from environment variables
+max_concurrent_requests_per_key = {}
+for key, value in os.environ.items():
+    if key.startswith("MAX_CONCURRENT_REQUESTS_PER_KEY_"):
+        provider = key.replace("MAX_CONCURRENT_REQUESTS_PER_KEY_", "").lower()
+        try:
+            max_concurrent = int(value)
+            if max_concurrent < 1:
+                logging.warning(
+                    f"Invalid max_concurrent value for provider '{provider}': {value}. Must be >= 1. Using default (1)."
+                )
+                max_concurrent = 1
+            max_concurrent_requests_per_key[provider] = max_concurrent
+            logging.debug(
+                f"Loaded max concurrent requests for provider '{provider}': {max_concurrent}"
+            )
+        except ValueError:
+            logging.warning(
+                f"Invalid max_concurrent value for provider '{provider}': {value}. Using default (1)."
+            )
+
+
+# --- Lifespan Management ---
+@asynccontextmanager
+async def lifespan(app: FastAPI):
+    """Manage the RotatingClient's lifecycle with the app's lifespan."""
+    # [MODIFIED] Perform skippable OAuth initialization at startup
+    skip_oauth_init = os.getenv("SKIP_OAUTH_INIT_CHECK", "false").lower() == "true"
+
+    # The CredentialManager now handles all discovery, including .env overrides.
+    # We pass all environment variables to it for this purpose.
+    cred_manager = CredentialManager(os.environ)
+    oauth_credentials = cred_manager.discover_and_prepare()
+
+    if not skip_oauth_init and oauth_credentials:
+        logging.info("Starting OAuth credential validation and deduplication...")
+        processed_emails = {}  # email -> {provider: path}
+        credentials_to_initialize = {}  # provider -> [paths]
+        final_oauth_credentials = {}
+
+        # --- Pass 1: Pre-initialization Scan & Deduplication ---
+        # logging.info("Pass 1: Scanning for existing metadata to find duplicates...")
+        for provider, paths in oauth_credentials.items():
+            if provider not in credentials_to_initialize:
+                credentials_to_initialize[provider] = []
+            for path in paths:
+                # Skip env-based credentials (virtual paths) - they don't have metadata files
+                if path.startswith("env://"):
+                    credentials_to_initialize[provider].append(path)
+                    continue
+
+                try:
+                    with open(path, "r") as f:
+                        data = json.load(f)
+                    metadata = data.get("_proxy_metadata", {})
+                    email = metadata.get("email")
+
+                    if email:
+                        if email not in processed_emails:
+                            processed_emails[email] = {}
+
+                        if provider in processed_emails[email]:
+                            original_path = processed_emails[email][provider]
+                            logging.warning(
+                                f"Duplicate for '{email}' on '{provider}' found in pre-scan: '{Path(path).name}'. Original: '{Path(original_path).name}'. Skipping."
+                            )
+                            continue
+                        else:
+                            processed_emails[email][provider] = path
+
+                    credentials_to_initialize[provider].append(path)
+
+                except (FileNotFoundError, json.JSONDecodeError) as e:
+                    logging.warning(
+                        f"Could not pre-read metadata from '{path}': {e}. Will process during initialization."
+                    )
+                    credentials_to_initialize[provider].append(path)
+
+        # --- Pass 2: Parallel Initialization of Filtered Credentials ---
+        # logging.info("Pass 2: Initializing unique credentials and performing final check...")
+        async def process_credential(provider: str, path: str, provider_instance):
+            """Process a single credential: initialize and fetch user info."""
+            try:
+                await provider_instance.initialize_token(path)
+
+                if not hasattr(provider_instance, "get_user_info"):
+                    return (provider, path, None, None)
+
+                user_info = await provider_instance.get_user_info(path)
+                email = user_info.get("email")
+                return (provider, path, email, None)
+
+            except Exception as e:
+                logging.error(
+                    f"Failed to process OAuth token for {provider} at '{path}': {e}"
+                )
+                return (provider, path, None, e)
+
+        # Collect all tasks for parallel execution
+        tasks = []
+        for provider, paths in credentials_to_initialize.items():
+            if not paths:
+                continue
+
+            provider_plugin_class = PROVIDER_PLUGINS.get(provider)
+            if not provider_plugin_class:
+                continue
+
+            provider_instance = provider_plugin_class()
+
+            for path in paths:
+                tasks.append(process_credential(provider, path, provider_instance))
+
+        # Execute all credential processing tasks in parallel
+        results = await asyncio.gather(*tasks, return_exceptions=True)
+
+        # --- Pass 3: Sequential Deduplication and Final Assembly ---
+        for result in results:
+            # Handle exceptions from gather
+            if isinstance(result, Exception):
+                logging.error(f"Credential processing raised exception: {result}")
+                continue
+
+            provider, path, email, error = result
+
+            # Skip if there was an error
+            if error:
+                continue
+
+            # If provider doesn't support get_user_info, add directly
+            if email is None:
+                if provider not in final_oauth_credentials:
+                    final_oauth_credentials[provider] = []
+                final_oauth_credentials[provider].append(path)
+                continue
+
+            # Handle empty email
+            if not email:
+                logging.warning(
+                    f"Could not retrieve email for '{path}'. Treating as unique."
+                )
+                if provider not in final_oauth_credentials:
+                    final_oauth_credentials[provider] = []
+                final_oauth_credentials[provider].append(path)
+                continue
+
+            # Deduplication check
+            if email not in processed_emails:
+                processed_emails[email] = {}
+
+            if (
+                provider in processed_emails[email]
+                and processed_emails[email][provider] != path
+            ):
+                original_path = processed_emails[email][provider]
+                logging.warning(
+                    f"Duplicate for '{email}' on '{provider}' found post-init: '{Path(path).name}'. Original: '{Path(original_path).name}'. Skipping."
+                )
+                continue
+            else:
+                processed_emails[email][provider] = path
+                if provider not in final_oauth_credentials:
+                    final_oauth_credentials[provider] = []
+                final_oauth_credentials[provider].append(path)
+
+                # Update metadata (skip for env-based credentials - they don't have files)
+                if not path.startswith("env://"):
+                    try:
+                        with open(path, "r+") as f:
+                            data = json.load(f)
+                            metadata = data.get("_proxy_metadata", {})
+                            metadata["email"] = email
+                            metadata["last_check_timestamp"] = time.time()
+                            data["_proxy_metadata"] = metadata
+                            f.seek(0)
+                            json.dump(data, f, indent=2)
+                            f.truncate()
+                    except Exception as e:
+                        logging.error(f"Failed to update metadata for '{path}': {e}")
+
+        logging.info("OAuth credential processing complete.")
+        oauth_credentials = final_oauth_credentials
+
+    # [NEW] Load provider-specific params
+    litellm_provider_params = {
+        "gemini_cli": {"project_id": os.getenv("GEMINI_CLI_PROJECT_ID")}
+    }
+
+    # Load global timeout from environment (default 30 seconds)
+    global_timeout = int(os.getenv("GLOBAL_TIMEOUT", "30"))
+
+    # The client now uses the root logger configuration
+    client = RotatingClient(
+        api_keys=api_keys,
+        oauth_credentials=oauth_credentials,  # Pass OAuth config
+        configure_logging=True,
+        global_timeout=global_timeout,
+        litellm_provider_params=litellm_provider_params,
+        ignore_models=ignore_models,
+        whitelist_models=whitelist_models,
+        enable_request_logging=ENABLE_REQUEST_LOGGING,
+        max_concurrent_requests_per_key=max_concurrent_requests_per_key,
+    )
+
+    # Log loaded credentials summary (compact, always visible for deployment verification)
+    # _api_summary = ', '.join([f"{p}:{len(c)}" for p, c in api_keys.items()]) if api_keys else "none"
+    # _oauth_summary = ', '.join([f"{p}:{len(c)}" for p, c in oauth_credentials.items()]) if oauth_credentials else "none"
+    # _total_summary = ', '.join([f"{p}:{len(c)}" for p, c in client.all_credentials.items()])
+    # print(f"🔑 Credentials loaded: {_total_summary} (API: {_api_summary} | OAuth: {_oauth_summary})")
+    client.background_refresher.start()  # Start the background task
+    app.state.rotating_client = client
+
+    # Warn if no provider credentials are configured
+    if not client.all_credentials:
+        logging.warning("=" * 70)
+        logging.warning("⚠️  NO PROVIDER CREDENTIALS CONFIGURED")
+        logging.warning("The proxy is running but cannot serve any LLM requests.")
+        logging.warning(
+            "Launch the credential tool to add API keys or OAuth credentials."
+        )
+        logging.warning("  • Executable: Run with --add-credential flag")
+        logging.warning("  • Source: python src/proxy_app/main.py --add-credential")
+        logging.warning("=" * 70)
+
+    os.environ["LITELLM_LOG"] = "ERROR"
+    litellm.set_verbose = False
+    litellm.drop_params = True
+    if USE_EMBEDDING_BATCHER:
+        batcher = EmbeddingBatcher(client=client)
+        app.state.embedding_batcher = batcher
+        logging.info("RotatingClient and EmbeddingBatcher initialized.")
+    else:
+        app.state.embedding_batcher = None
+        logging.info("RotatingClient initialized (EmbeddingBatcher disabled).")
+
+    # Start model info service in background (fetches pricing/capabilities data)
+    # This runs asynchronously and doesn't block proxy startup
+    model_info_service = await init_model_info_service()
+    app.state.model_info_service = model_info_service
+    logging.info("Model info service started (fetching pricing data in background).")
+
+    yield
+
+    await client.background_refresher.stop()  # Stop the background task on shutdown
+    if app.state.embedding_batcher:
+        await app.state.embedding_batcher.stop()
+    await client.close()
+
+    # Stop model info service
+    if hasattr(app.state, "model_info_service") and app.state.model_info_service:
+        await app.state.model_info_service.stop()
+
+    if app.state.embedding_batcher:
+        logging.info("RotatingClient and EmbeddingBatcher closed.")
+    else:
+        logging.info("RotatingClient closed.")
+
+
+# --- FastAPI App Setup ---
+app = FastAPI(lifespan=lifespan)
+
+# Add CORS middleware to allow all origins, methods, and headers
+app.add_middleware(
+    CORSMiddleware,
+    allow_origins=["*"],  # Allows all origins
+    allow_credentials=True,
+    allow_methods=["*"],  # Allows all methods
+    allow_headers=["*"],  # Allows all headers
+)
+api_key_header = APIKeyHeader(name="Authorization", auto_error=False)
+
+
+def get_rotating_client(request: Request) -> RotatingClient:
+    """Dependency to get the rotating client instance from the app state."""
+    return request.app.state.rotating_client
+
+
+def get_embedding_batcher(request: Request) -> EmbeddingBatcher:
+    """Dependency to get the embedding batcher instance from the app state."""
+    return request.app.state.embedding_batcher
+
+
+async def verify_api_key(auth: str = Depends(api_key_header)):
+    """Dependency to verify the proxy API key."""
+    # If PROXY_API_KEY is not set or empty, skip verification (open access)
+    if not PROXY_API_KEY:
+        return auth
+    if not auth or auth != f"Bearer {PROXY_API_KEY}":
+        raise HTTPException(status_code=401, detail="Invalid or missing API Key")
+    return auth
+
+
+# --- Anthropic API Key Header ---
+anthropic_api_key_header = APIKeyHeader(name="x-api-key", auto_error=False)
+
+
+async def verify_anthropic_api_key(
+    x_api_key: str = Depends(anthropic_api_key_header),
+    auth: str = Depends(api_key_header),
+):
+    """
+    Dependency to verify API key for Anthropic endpoints.
+    Accepts either x-api-key header (Anthropic style) or Authorization Bearer (OpenAI style).
+    """
+    # Check x-api-key first (Anthropic style)
+    if x_api_key and x_api_key == PROXY_API_KEY:
+        return x_api_key
+    # Fall back to Bearer token (OpenAI style)
+    if auth and auth == f"Bearer {PROXY_API_KEY}":
+        return auth
+    raise HTTPException(status_code=401, detail="Invalid or missing API Key")
+
+
+async def streaming_response_wrapper(
+    request: Request,
+    request_data: dict,
+    response_stream: AsyncGenerator[str, None],
+    logger: Optional[RawIOLogger] = None,
+) -> AsyncGenerator[str, None]:
+    """
+    Wraps a streaming response to log the full response after completion
+    and ensures any errors during the stream are sent to the client.
+    """
+    response_chunks = []
+    full_response = {}
+
+    try:
+        async for chunk_str in response_stream:
+            if await request.is_disconnected():
+                logging.warning("Client disconnected, stopping stream.")
+                break
+            yield chunk_str
+            if chunk_str.strip() and chunk_str.startswith("data:"):
+                content = chunk_str[len("data:") :].strip()
+                if content != "[DONE]":
+                    try:
+                        chunk_data = json.loads(content)
+                        response_chunks.append(chunk_data)
+                        if logger:
+                            logger.log_stream_chunk(chunk_data)
+                    except json.JSONDecodeError:
+                        pass
+    except Exception as e:
+        logging.error(f"An error occurred during the response stream: {e}")
+        # Yield a final error message to the client to ensure they are not left hanging.
+        error_payload = {
+            "error": {
+                "message": f"An unexpected error occurred during the stream: {str(e)}",
+                "type": "proxy_internal_error",
+                "code": 500,
+            }
+        }
+        yield f"data: {json.dumps(error_payload)}\n\n"
+        yield "data: [DONE]\n\n"
+        # Also log this as a failed request
+        if logger:
+            logger.log_final_response(
+                status_code=500, headers=None, body={"error": str(e)}
+            )
+        return  # Stop further processing
+    finally:
+        if response_chunks:
+            # --- Aggregation Logic ---
+            final_message = {"role": "assistant"}
+            aggregated_tool_calls = {}
+            usage_data = None
+            finish_reason = None
+
+            for chunk in response_chunks:
+                if "choices" in chunk and chunk["choices"]:
+                    choice = chunk["choices"][0]
+                    delta = choice.get("delta", {})
+
+                    # Dynamically aggregate all fields from the delta
+                    for key, value in delta.items():
+                        if value is None:
+                            continue
+
+                        if key == "content":
+                            if "content" not in final_message:
+                                final_message["content"] = ""
+                            if value:
+                                final_message["content"] += value
+
+                        elif key == "tool_calls":
+                            for tc_chunk in value:
+                                index = tc_chunk["index"]
+                                if index not in aggregated_tool_calls:
+                                    aggregated_tool_calls[index] = {
+                                        "type": "function",
+                                        "function": {"name": "", "arguments": ""},
+                                    }
+                                # Ensure 'function' key exists for this index before accessing its sub-keys
+                                if "function" not in aggregated_tool_calls[index]:
+                                    aggregated_tool_calls[index]["function"] = {
+                                        "name": "",
+                                        "arguments": "",
+                                    }
+                                if tc_chunk.get("id"):
+                                    aggregated_tool_calls[index]["id"] = tc_chunk["id"]
+                                if "function" in tc_chunk:
+                                    if "name" in tc_chunk["function"]:
+                                        if tc_chunk["function"]["name"] is not None:
+                                            aggregated_tool_calls[index]["function"][
+                                                "name"
+                                            ] += tc_chunk["function"]["name"]
+                                    if "arguments" in tc_chunk["function"]:
+                                        if (
+                                            tc_chunk["function"]["arguments"]
+                                            is not None
+                                        ):
+                                            aggregated_tool_calls[index]["function"][
+                                                "arguments"
+                                            ] += tc_chunk["function"]["arguments"]
+
+                        elif key == "function_call":
+                            if "function_call" not in final_message:
+                                final_message["function_call"] = {
+                                    "name": "",
+                                    "arguments": "",
+                                }
+                            if "name" in value:
+                                if value["name"] is not None:
+                                    final_message["function_call"]["name"] += value[
+                                        "name"
+                                    ]
+                            if "arguments" in value:
+                                if value["arguments"] is not None:
+                                    final_message["function_call"]["arguments"] += (
+                                        value["arguments"]
+                                    )
+
+                        else:  # Generic key handling for other data like 'reasoning'
+                            # FIX: Role should always replace, never concatenate
+                            if key == "role":
+                                final_message[key] = value
+                            elif key not in final_message:
+                                final_message[key] = value
+                            elif isinstance(final_message.get(key), str):
+                                final_message[key] += value
+                            else:
+                                final_message[key] = value
+
+                    if "finish_reason" in choice and choice["finish_reason"]:
+                        finish_reason = choice["finish_reason"]
+
+                if "usage" in chunk and chunk["usage"]:
+                    usage_data = chunk["usage"]
+
+            # --- Final Response Construction ---
+            if aggregated_tool_calls:
+                final_message["tool_calls"] = list(aggregated_tool_calls.values())
+                # CRITICAL FIX: Override finish_reason when tool_calls exist
+                # This ensures OpenCode and other agentic systems continue the conversation loop
+                finish_reason = "tool_calls"
+
+            # Ensure standard fields are present for consistent logging
+            for field in ["content", "tool_calls", "function_call"]:
+                if field not in final_message:
+                    final_message[field] = None
+
+            first_chunk = response_chunks[0]
+            final_choice = {
+                "index": 0,
+                "message": final_message,
+                "finish_reason": finish_reason,
+            }
+
+            full_response = {
+                "id": first_chunk.get("id"),
+                "object": "chat.completion",
+                "created": first_chunk.get("created"),
+                "model": first_chunk.get("model"),
+                "choices": [final_choice],
+                "usage": usage_data,
+            }
+
+        if logger:
+            logger.log_final_response(
+                status_code=200,
+                headers=None,  # Headers are not available at this stage
+                body=full_response,
+            )
+
+
+@app.post("/v1/chat/completions")
+async def chat_completions(
+    request: Request,
+    client: RotatingClient = Depends(get_rotating_client),
+    _=Depends(verify_api_key),
+):
+    """
+    OpenAI-compatible endpoint powered by the RotatingClient.
+    Handles both streaming and non-streaming responses and logs them.
+    """
+    # Raw I/O logger captures unmodified HTTP data at proxy boundary (disabled by default)
+    raw_logger = RawIOLogger() if ENABLE_RAW_LOGGING else None
+    try:
+        # Read and parse the request body only once at the beginning.
+        try:
+            request_data = await request.json()
+        except json.JSONDecodeError:
+            raise HTTPException(status_code=400, detail="Invalid JSON in request body.")
+
+        # Global temperature=0 override (controlled by .env variable, default: OFF)
+        # Low temperature makes models deterministic and prone to following training data
+        # instead of actual schemas, which can cause tool hallucination
+        # Modes: "remove" = delete temperature key, "set" = change to 1.0, "false" = disabled
+        override_temp_zero = os.getenv("OVERRIDE_TEMPERATURE_ZERO", "false").lower()
+
+        if (
+            override_temp_zero in ("remove", "set", "true", "1", "yes")
+            and "temperature" in request_data
+            and request_data["temperature"] == 0
+        ):
+            if override_temp_zero == "remove":
+                # Remove temperature key entirely
+                del request_data["temperature"]
+                logging.debug(
+                    "OVERRIDE_TEMPERATURE_ZERO=remove: Removed temperature=0 from request"
+                )
+            else:
+                # Set to 1.0 (for "set", "true", "1", "yes")
+                request_data["temperature"] = 1.0
+                logging.debug(
+                    "OVERRIDE_TEMPERATURE_ZERO=set: Converting temperature=0 to temperature=1.0"
+                )
+
+        # If raw logging is enabled, capture the unmodified request data.
+        if raw_logger:
+            raw_logger.log_request(headers=request.headers, body=request_data)
+
+        # Extract and log specific reasoning parameters for monitoring.
+        model = request_data.get("model")
+        generation_cfg = (
+            request_data.get("generationConfig", {})
+            or request_data.get("generation_config", {})
+            or {}
+        )
+        reasoning_effort = request_data.get("reasoning_effort") or generation_cfg.get(
+            "reasoning_effort"
+        )
+
+        logging.getLogger("rotator_library").debug(
+            f"Handling reasoning parameters: model={model}, reasoning_effort={reasoning_effort}"
+        )
+
+        # Log basic request info to console (this is a separate, simpler logger).
+        log_request_to_console(
+            url=str(request.url),
+            headers=dict(request.headers),
+            client_info=(request.client.host, request.client.port),
+            request_data=request_data,
+        )
+        is_streaming = request_data.get("stream", False)
+
+        if is_streaming:
+            response_generator = client.acompletion(request=request, **request_data)
+            return StreamingResponse(
+                streaming_response_wrapper(
+                    request, request_data, response_generator, raw_logger
+                ),
+                media_type="text/event-stream",
+            )
+        else:
+            response = await client.acompletion(request=request, **request_data)
+            if raw_logger:
+                # Assuming response has status_code and headers attributes
+                # This might need adjustment based on the actual response object
+                response_headers = (
+                    response.headers if hasattr(response, "headers") else None
+                )
+                status_code = (
+                    response.status_code if hasattr(response, "status_code") else 200
+                )
+                raw_logger.log_final_response(
+                    status_code=status_code,
+                    headers=response_headers,
+                    body=response.model_dump(),
+                )
+            return response
+
+    except (
+        litellm.InvalidRequestError,
+        ValueError,
+        litellm.ContextWindowExceededError,
+    ) as e:
+        raise HTTPException(status_code=400, detail=f"Invalid Request: {str(e)}")
+    except litellm.AuthenticationError as e:
+        raise HTTPException(status_code=401, detail=f"Authentication Error: {str(e)}")
+    except litellm.RateLimitError as e:
+        raise HTTPException(status_code=429, detail=f"Rate Limit Exceeded: {str(e)}")
+    except (litellm.ServiceUnavailableError, litellm.APIConnectionError) as e:
+        raise HTTPException(status_code=503, detail=f"Service Unavailable: {str(e)}")
+    except litellm.Timeout as e:
+        raise HTTPException(status_code=504, detail=f"Gateway Timeout: {str(e)}")
+    except (litellm.InternalServerError, litellm.OpenAIError) as e:
+        raise HTTPException(status_code=502, detail=f"Bad Gateway: {str(e)}")
+    except Exception as e:
+        logging.error(f"Request failed after all retries: {e}")
+        # Optionally log the failed request
+        if ENABLE_REQUEST_LOGGING:
+            try:
+                request_data = await request.json()
+            except json.JSONDecodeError:
+                request_data = {"error": "Could not parse request body"}
+            if raw_logger:
+                raw_logger.log_final_response(
+                    status_code=500, headers=None, body={"error": str(e)}
+                )
+        raise HTTPException(status_code=500, detail=str(e))
+
+
+# --- Anthropic Messages API Endpoint ---
+@app.post("/v1/messages")
+async def anthropic_messages(
+    request: Request,
+    body: AnthropicMessagesRequest,
+    client: RotatingClient = Depends(get_rotating_client),
+    _=Depends(verify_anthropic_api_key),
+):
+    """
+    Anthropic-compatible Messages API endpoint.
+
+    Accepts requests in Anthropic's format and returns responses in Anthropic's format.
+    Internally translates to OpenAI format for processing via LiteLLM.
+
+    This endpoint is compatible with Claude Code and other Anthropic API clients.
+    """
+    # Initialize raw I/O logger if enabled (for debugging proxy boundary)
+    logger = RawIOLogger() if ENABLE_RAW_LOGGING else None
+
+    # Log raw Anthropic request if raw logging is enabled
+    if logger:
+        logger.log_request(
+            headers=dict(request.headers),
+            body=body.model_dump(exclude_none=True),
+        )
+
+    try:
+        # Log the request to console
+        log_request_to_console(
+            url=str(request.url),
+            headers=dict(request.headers),
+            client_info=(
+                request.client.host if request.client else "unknown",
+                request.client.port if request.client else 0,
+            ),
+            request_data=body.model_dump(exclude_none=True),
+        )
+
+        # Use the library method to handle the request
+        result = await client.anthropic_messages(body, raw_request=request)
+
+        if body.stream:
+            # Streaming response
+            return StreamingResponse(
+                result,
+                media_type="text/event-stream",
+                headers={
+                    "Cache-Control": "no-cache",
+                    "Connection": "keep-alive",
+                    "X-Accel-Buffering": "no",
+                },
+            )
+        else:
+            # Non-streaming response
+            if logger:
+                logger.log_final_response(
+                    status_code=200,
+                    headers=None,
+                    body=result,
+                )
+            return JSONResponse(content=result)
+
+    except (
+        litellm.InvalidRequestError,
+        ValueError,
+        litellm.ContextWindowExceededError,
+    ) as e:
+        error_response = {
+            "type": "error",
+            "error": {"type": "invalid_request_error", "message": str(e)},
+        }
+        raise HTTPException(status_code=400, detail=error_response)
+    except litellm.AuthenticationError as e:
+        error_response = {
+            "type": "error",
+            "error": {"type": "authentication_error", "message": str(e)},
+        }
+        raise HTTPException(status_code=401, detail=error_response)
+    except litellm.RateLimitError as e:
+        error_response = {
+            "type": "error",
+            "error": {"type": "rate_limit_error", "message": str(e)},
+        }
+        raise HTTPException(status_code=429, detail=error_response)
+    except (litellm.ServiceUnavailableError, litellm.APIConnectionError) as e:
+        error_response = {
+            "type": "error",
+            "error": {"type": "api_error", "message": str(e)},
+        }
+        raise HTTPException(status_code=503, detail=error_response)
+    except litellm.Timeout as e:
+        error_response = {
+            "type": "error",
+            "error": {"type": "api_error", "message": f"Request timed out: {str(e)}"},
+        }
+        raise HTTPException(status_code=504, detail=error_response)
+    except Exception as e:
+        logging.error(f"Anthropic messages endpoint error: {e}")
+        if logger:
+            logger.log_final_response(
+                status_code=500,
+                headers=None,
+                body={"error": str(e)},
+            )
+        error_response = {
+            "type": "error",
+            "error": {"type": "api_error", "message": str(e)},
+        }
+        raise HTTPException(status_code=500, detail=error_response)
+
+
+# --- Anthropic Count Tokens Endpoint ---
+@app.post("/v1/messages/count_tokens")
+async def anthropic_count_tokens(
+    request: Request,
+    body: AnthropicCountTokensRequest,
+    client: RotatingClient = Depends(get_rotating_client),
+    _=Depends(verify_anthropic_api_key),
+):
+    """
+    Anthropic-compatible count_tokens endpoint.
+
+    Counts the number of tokens that would be used by a Messages API request.
+    This is useful for estimating costs and managing context windows.
+
+    Accepts requests in Anthropic's format and returns token count in Anthropic's format.
+    """
+    try:
+        # Use the library method to handle the request
+        result = await client.anthropic_count_tokens(body)
+        return JSONResponse(content=result)
+
+    except (
+        litellm.InvalidRequestError,
+        ValueError,
+        litellm.ContextWindowExceededError,
+    ) as e:
+        error_response = {
+            "type": "error",
+            "error": {"type": "invalid_request_error", "message": str(e)},
+        }
+        raise HTTPException(status_code=400, detail=error_response)
+    except litellm.AuthenticationError as e:
+        error_response = {
+            "type": "error",
+            "error": {"type": "authentication_error", "message": str(e)},
+        }
+        raise HTTPException(status_code=401, detail=error_response)
+    except Exception as e:
+        logging.error(f"Anthropic count_tokens endpoint error: {e}")
+        error_response = {
+            "type": "error",
+            "error": {"type": "api_error", "message": str(e)},
+        }
+        raise HTTPException(status_code=500, detail=error_response)
+
+
+@app.post("/v1/embeddings")
+async def embeddings(
+    request: Request,
+    body: EmbeddingRequest,
+    client: RotatingClient = Depends(get_rotating_client),
+    batcher: Optional[EmbeddingBatcher] = Depends(get_embedding_batcher),
+    _=Depends(verify_api_key),
+):
+    """
+    OpenAI-compatible endpoint for creating embeddings.
+    Supports two modes based on the USE_EMBEDDING_BATCHER flag:
+    - True: Uses a server-side batcher for high throughput.
+    - False: Passes requests directly to the provider.
+    """
+    try:
+        request_data = body.model_dump(exclude_none=True)
+        log_request_to_console(
+            url=str(request.url),
+            headers=dict(request.headers),
+            client_info=(request.client.host, request.client.port),
+            request_data=request_data,
+        )
+        if USE_EMBEDDING_BATCHER and batcher:
+            # --- Server-Side Batching Logic ---
+            request_data = body.model_dump(exclude_none=True)
+            inputs = request_data.get("input", [])
+            if isinstance(inputs, str):
+                inputs = [inputs]
+
+            tasks = []
+            for single_input in inputs:
+                individual_request = request_data.copy()
+                individual_request["input"] = single_input
+                tasks.append(batcher.add_request(individual_request))
+
+            results = await asyncio.gather(*tasks)
+
+            all_data = []
+            total_prompt_tokens = 0
+            total_tokens = 0
+            for i, result in enumerate(results):
+                result["data"][0]["index"] = i
+                all_data.extend(result["data"])
+                total_prompt_tokens += result["usage"]["prompt_tokens"]
+                total_tokens += result["usage"]["total_tokens"]
+
+            final_response_data = {
+                "object": "list",
+                "model": results[0]["model"],
+                "data": all_data,
+                "usage": {
+                    "prompt_tokens": total_prompt_tokens,
+                    "total_tokens": total_tokens,
+                },
+            }
+            response = litellm.EmbeddingResponse(**final_response_data)
+
+        else:
+            # --- Direct Pass-Through Logic ---
+            request_data = body.model_dump(exclude_none=True)
+            if isinstance(request_data.get("input"), str):
+                request_data["input"] = [request_data["input"]]
+
+            response = await client.aembedding(request=request, **request_data)
+
+        return response
+
+    except HTTPException as e:
+        # Re-raise HTTPException to ensure it's not caught by the generic Exception handler
+        raise e
+    except (
+        litellm.InvalidRequestError,
+        ValueError,
+        litellm.ContextWindowExceededError,
+    ) as e:
+        raise HTTPException(status_code=400, detail=f"Invalid Request: {str(e)}")
+    except litellm.AuthenticationError as e:
+        raise HTTPException(status_code=401, detail=f"Authentication Error: {str(e)}")
+    except litellm.RateLimitError as e:
+        raise HTTPException(status_code=429, detail=f"Rate Limit Exceeded: {str(e)}")
+    except (litellm.ServiceUnavailableError, litellm.APIConnectionError) as e:
+        raise HTTPException(status_code=503, detail=f"Service Unavailable: {str(e)}")
+    except litellm.Timeout as e:
+        raise HTTPException(status_code=504, detail=f"Gateway Timeout: {str(e)}")
+    except (litellm.InternalServerError, litellm.OpenAIError) as e:
+        raise HTTPException(status_code=502, detail=f"Bad Gateway: {str(e)}")
+    except Exception as e:
+        logging.error(f"Embedding request failed: {e}")
+        raise HTTPException(status_code=500, detail=str(e))
+
+
+@app.get("/")
+def read_root():
+    return {"Status": "API Key Proxy is running"}
+
+
+@app.get("/v1/models")
+async def list_models(
+    request: Request,
+    client: RotatingClient = Depends(get_rotating_client),
+    _=Depends(verify_api_key),
+    enriched: bool = True,
+):
+    """
+    Returns a list of available models in the OpenAI-compatible format.
+
+    Query Parameters:
+        enriched: If True (default), returns detailed model info with pricing and capabilities.
+                  If False, returns minimal OpenAI-compatible response.
+    """
+    model_ids = await client.get_all_available_models(grouped=False)
+
+    if enriched and hasattr(request.app.state, "model_info_service"):
+        model_info_service = request.app.state.model_info_service
+        if model_info_service.is_ready:
+            # Return enriched model data
+            enriched_data = model_info_service.enrich_model_list(model_ids)
+            return {"object": "list", "data": enriched_data}
+
+    # Fallback to basic model cards
+    model_cards = [
+        {
+            "id": model_id,
+            "object": "model",
+            "created": int(time.time()),
+            "owned_by": "Mirro-Proxy",
+        }
+        for model_id in model_ids
+    ]
+    return {"object": "list", "data": model_cards}
+
+
+@app.get("/v1/models/{model_id:path}")
+async def get_model(
+    model_id: str,
+    request: Request,
+    _=Depends(verify_api_key),
+):
+    """
+    Returns detailed information about a specific model.
+
+    Path Parameters:
+        model_id: The model ID (e.g., "anthropic/claude-3-opus", "openrouter/openai/gpt-4")
+    """
+    if hasattr(request.app.state, "model_info_service"):
+        model_info_service = request.app.state.model_info_service
+        if model_info_service.is_ready:
+            info = model_info_service.get_model_info(model_id)
+            if info:
+                return info.to_dict()
+
+    # Return basic info if service not ready or model not found
+    return {
+        "id": model_id,
+        "object": "model",
+        "created": int(time.time()),
+        "owned_by": model_id.split("/")[0] if "/" in model_id else "unknown",
+    }
+
+
+@app.get("/v1/model-info/stats")
+async def model_info_stats(
+    request: Request,
+    _=Depends(verify_api_key),
+):
+    """
+    Returns statistics about the model info service (for monitoring/debugging).
+    """
+    if hasattr(request.app.state, "model_info_service"):
+        return request.app.state.model_info_service.get_stats()
+    return {"error": "Model info service not initialized"}
+
+
+@app.get("/v1/providers")
+async def list_providers(_=Depends(verify_api_key)):
+    """
+    Returns a list of all available providers.
+    """
+    return list(PROVIDER_PLUGINS.keys())
+
+
+@app.get("/v1/quota-stats")
+async def get_quota_stats(
+    request: Request,
+    client: RotatingClient = Depends(get_rotating_client),
+    _=Depends(verify_api_key),
+    provider: str = None,
+):
+    """
+    Returns quota and usage statistics for all credentials.
+
+    This returns cached data from the proxy without making external API calls.
+    Use POST to reload from disk or force refresh from external APIs.
+
+    Query Parameters:
+        provider: Optional filter to return stats for a specific provider only
+
+    Returns:
+        {
+            "providers": {
+                "provider_name": {
+                    "credential_count": int,
+                    "active_count": int,
+                    "on_cooldown_count": int,
+                    "exhausted_count": int,
+                    "total_requests": int,
+                    "tokens": {...},
+                    "approx_cost": float | null,
+                    "quota_groups": {...},  // For Antigravity
+                    "credentials": [...]
+                }
+            },
+            "summary": {...},
+            "data_source": "cache",
+            "timestamp": float
+        }
+    """
+    try:
+        stats = await client.get_quota_stats(provider_filter=provider)
+        return stats
+    except Exception as e:
+        logging.error(f"Failed to get quota stats: {e}")
+        raise HTTPException(status_code=500, detail=str(e))
+
+
+@app.post("/v1/quota-stats")
+async def refresh_quota_stats(
+    request: Request,
+    client: RotatingClient = Depends(get_rotating_client),
+    _=Depends(verify_api_key),
+):
+    """
+    Refresh quota and usage statistics.
+
+    Request body:
+        {
+            "action": "reload" | "force_refresh",
+            "scope": "all" | "provider" | "credential",
+            "provider": "antigravity",  // required if scope != "all"
+            "credential": "antigravity_oauth_1.json"  // required if scope == "credential"
+        }
+
+    Actions:
+        - reload: Re-read data from disk (no external API calls)
+        - force_refresh: For Antigravity, fetch live quota from API.
+                        For other providers, same as reload.
+
+    Returns:
+        Same as GET, plus a "refresh_result" field with operation details.
+    """
+    try:
+        data = await request.json()
+        action = data.get("action", "reload")
+        scope = data.get("scope", "all")
+        provider = data.get("provider")
+        credential = data.get("credential")
+
+        # Validate parameters
+        if action not in ("reload", "force_refresh"):
+            raise HTTPException(
+                status_code=400,
+                detail="action must be 'reload' or 'force_refresh'",
+            )
+
+        if scope not in ("all", "provider", "credential"):
+            raise HTTPException(
+                status_code=400,
+                detail="scope must be 'all', 'provider', or 'credential'",
+            )
+
+        if scope in ("provider", "credential") and not provider:
+            raise HTTPException(
+                status_code=400,
+                detail="'provider' is required when scope is 'provider' or 'credential'",
+            )
+
+        if scope == "credential" and not credential:
+            raise HTTPException(
+                status_code=400,
+                detail="'credential' is required when scope is 'credential'",
+            )
+
+        refresh_result = {
+            "action": action,
+            "scope": scope,
+            "provider": provider,
+            "credential": credential,
+        }
+
+        if action == "reload":
+            # Just reload from disk
+            start_time = time.time()
+            await client.reload_usage_from_disk()
+            refresh_result["duration_ms"] = int((time.time() - start_time) * 1000)
+            refresh_result["success"] = True
+            refresh_result["message"] = "Reloaded usage data from disk"
+
+        elif action == "force_refresh":
+            # Force refresh from external API (for supported providers like Antigravity)
+            result = await client.force_refresh_quota(
+                provider=provider if scope in ("provider", "credential") else None,
+                credential=credential if scope == "credential" else None,
+            )
+            refresh_result.update(result)
+            refresh_result["success"] = result["failed_count"] == 0
+
+        # Get updated stats
+        stats = await client.get_quota_stats(provider_filter=provider)
+        stats["refresh_result"] = refresh_result
+        stats["data_source"] = "refreshed"
+
+        return stats
+
+    except HTTPException:
+        raise
+    except Exception as e:
+        logging.error(f"Failed to refresh quota stats: {e}")
+        raise HTTPException(status_code=500, detail=str(e))
+
+
+@app.post("/v1/token-count")
+async def token_count(
+    request: Request,
+    client: RotatingClient = Depends(get_rotating_client),
+    _=Depends(verify_api_key),
+):
+    """
+    Calculates the token count for a given list of messages and a model.
+    """
+    try:
+        data = await request.json()
+        model = data.get("model")
+        messages = data.get("messages")
+
+        if not model or not messages:
+            raise HTTPException(
+                status_code=400, detail="'model' and 'messages' are required."
+            )
+
+        count = client.token_count(**data)
+        return {"token_count": count}
+
+    except Exception as e:
+        logging.error(f"Token count failed: {e}")
+        raise HTTPException(status_code=500, detail=str(e))
+
+
+@app.post("/v1/cost-estimate")
+async def cost_estimate(request: Request, _=Depends(verify_api_key)):
+    """
+    Estimates the cost for a request based on token counts and model pricing.
+
+    Request body:
+        {
+            "model": "anthropic/claude-3-opus",
+            "prompt_tokens": 1000,
+            "completion_tokens": 500,
+            "cache_read_tokens": 0,       # optional
+            "cache_creation_tokens": 0    # optional
+        }
+
+    Returns:
+        {
+            "model": "anthropic/claude-3-opus",
+            "cost": 0.0375,
+            "currency": "USD",
+            "pricing": {
+                "input_cost_per_token": 0.000015,
+                "output_cost_per_token": 0.000075
+            },
+            "source": "model_info_service"  # or "litellm_fallback"
+        }
+    """
+    try:
+        data = await request.json()
+        model = data.get("model")
+        prompt_tokens = data.get("prompt_tokens", 0)
+        completion_tokens = data.get("completion_tokens", 0)
+        cache_read_tokens = data.get("cache_read_tokens", 0)
+        cache_creation_tokens = data.get("cache_creation_tokens", 0)
+
+        if not model:
+            raise HTTPException(status_code=400, detail="'model' is required.")
+
+        result = {
+            "model": model,
+            "cost": None,
+            "currency": "USD",
+            "pricing": {},
+            "source": None,
+        }
+
+        # Try model info service first
+        if hasattr(request.app.state, "model_info_service"):
+            model_info_service = request.app.state.model_info_service
+            if model_info_service.is_ready:
+                cost = model_info_service.calculate_cost(
+                    model,
+                    prompt_tokens,
+                    completion_tokens,
+                    cache_read_tokens,
+                    cache_creation_tokens,
+                )
+                if cost is not None:
+                    cost_info = model_info_service.get_cost_info(model)
+                    result["cost"] = cost
+                    result["pricing"] = cost_info or {}
+                    result["source"] = "model_info_service"
+                    return result
+
+        # Fallback to litellm
+        try:
+            import litellm
+
+            # Create a mock response for cost calculation
+            model_info = litellm.get_model_info(model)
+            input_cost = model_info.get("input_cost_per_token", 0)
+            output_cost = model_info.get("output_cost_per_token", 0)
+
+            if input_cost or output_cost:
+                cost = (prompt_tokens * input_cost) + (completion_tokens * output_cost)
+                result["cost"] = cost
+                result["pricing"] = {
+                    "input_cost_per_token": input_cost,
+                    "output_cost_per_token": output_cost,
+                }
+                result["source"] = "litellm_fallback"
+                return result
+        except Exception:
+            pass
+
+        result["source"] = "unknown"
+        result["error"] = "Pricing data not available for this model"
+        return result
+
+    except HTTPException:
+        raise
+    except Exception as e:
+        logging.error(f"Cost estimate failed: {e}")
+        raise HTTPException(status_code=500, detail=str(e))
+
+
+if __name__ == "__main__":
+    # Define ENV_FILE for onboarding checks using centralized path
+    ENV_FILE = get_data_file(".env")
+
+    # Check if launcher TUI should be shown (no arguments provided)
+    if len(sys.argv) == 1:
+        # No arguments - show launcher TUI (lazy import)
+        from proxy_app.launcher_tui import run_launcher_tui
+
+        run_launcher_tui()
+        # Launcher modifies sys.argv and returns, or exits if user chose Exit
+        # If we get here, user chose "Run Proxy" and sys.argv is modified
+        # Re-parse arguments with modified sys.argv
+        args = parser.parse_args()
+
+    def needs_onboarding() -> bool:
+        """
+        Check if the proxy needs onboarding (first-time setup).
+        Returns True if onboarding is needed, False otherwise.
+        """
+        # Only check if .env file exists
+        # PROXY_API_KEY is optional (will show warning if not set)
+        if not ENV_FILE.is_file():
+            return True
+
+        return False
+
+    def show_onboarding_message():
+        """Display clear explanatory message for why onboarding is needed."""
+        os.system(
+            "cls" if os.name == "nt" else "clear"
+        )  # Clear terminal for clean presentation
+        console.print(
+            Panel.fit(
+                "[bold cyan]🚀 LLM API Key Proxy - First Time Setup[/bold cyan]",
+                border_style="cyan",
+            )
+        )
+        console.print("[bold yellow]⚠️  Configuration Required[/bold yellow]\n")
+
+        console.print("The proxy needs initial configuration:")
+        console.print("  [red]❌ No .env file found[/red]")
+
+        console.print("\n[bold]Why this matters:[/bold]")
+        console.print("  • The .env file stores your credentials and settings")
+        console.print("  • PROXY_API_KEY protects your proxy from unauthorized access")
+        console.print("  • Provider API keys enable LLM access")
+
+        console.print("\n[bold]What happens next:[/bold]")
+        console.print("  1. We'll create a .env file with PROXY_API_KEY")
+        console.print("  2. You can add LLM provider credentials (API keys or OAuth)")
+        console.print("  3. The proxy will then start normally")
+
+        console.print(
+            "\n[bold yellow]⚠️  Note:[/bold yellow] The credential tool adds PROXY_API_KEY by default."
+        )
+        console.print("   You can remove it later if you want an unsecured proxy.\n")
+
+        console.input(
+            "[bold green]Press Enter to launch the credential setup tool...[/bold green]"
+        )
+
+    # Check if user explicitly wants to add credentials
+    if args.add_credential:
+        # Import and call ensure_env_defaults to create .env and PROXY_API_KEY if needed
+        from rotator_library.credential_tool import ensure_env_defaults
+
+        ensure_env_defaults()
+        # Reload environment variables after ensure_env_defaults creates/updates .env
+        load_dotenv(ENV_FILE, override=True)
+        run_credential_tool()
+    else:
+        # Check if onboarding is needed
+        if needs_onboarding():
+            # Import console from rich for better messaging
+            from rich.console import Console
+            from rich.panel import Panel
+
+            console = Console()
+
+            # Show clear explanatory message
+            show_onboarding_message()
+
+            # Launch credential tool automatically
+            from rotator_library.credential_tool import ensure_env_defaults
+
+            ensure_env_defaults()
+            load_dotenv(ENV_FILE, override=True)
+            run_credential_tool()
+
+            # After credential tool exits, reload and re-check
+            load_dotenv(ENV_FILE, override=True)
+            # Re-read PROXY_API_KEY from environment
+            PROXY_API_KEY = os.getenv("PROXY_API_KEY")
+
+            # Verify onboarding is complete
+            if needs_onboarding():
+                console.print("\n[bold red]❌ Configuration incomplete.[/bold red]")
+                console.print(
+                    "The proxy still cannot start. Please ensure PROXY_API_KEY is set in .env\n"
+                )
+                sys.exit(1)
+            else:
+                console.print("\n[bold green]✅ Configuration complete![/bold green]")
+                console.print("\nStarting proxy server...\n")
+
+        import uvicorn
+
+        uvicorn.run(app, host=args.host, port=args.port)

src/proxy_app/provider_urls.py

@@ -0,0 +1,76 @@
+# SPDX-License-Identifier: MIT
+# Copyright (c) 2026 Mirrowel
+
+import os
+from typing import Optional
+
+# A comprehensive map of provider names to their base URLs.
+PROVIDER_URL_MAP = {
+    "perplexity": "https://api.perplexity.ai",
+    "anyscale": "https://api.endpoints.anyscale.com/v1",
+    "deepinfra": "https://api.deepinfra.com/v1/openai",
+    "mistral": "https://api.mistral.ai/v1",
+    "groq": "https://api.groq.com/openai/v1",
+    "nvidia_nim": "https://integrate.api.nvidia.com/v1",
+    "cerebras": "https://api.cerebras.ai/v1",
+    "sambanova": "https://api.sambanova.ai/v1",
+    "ai21_chat": "https://api.ai21.com/studio/v1",
+    "codestral": "https://codestral.mistral.ai/v1",
+    "text-completion-codestral": "https://codestral.mistral.ai/v1",
+    "empower": "https://app.empower.dev/api/v1",
+    "deepseek": "https://api.deepseek.com/v1",
+    "friendliai": "https://api.friendli.ai/serverless/v1",
+    "galadriel": "https://api.galadriel.com/v1",
+    "meta_llama": "https://api.llama.com/compat/v1",
+    "featherless_ai": "https://api.featherless.ai/v1",
+    "nscale": "https://api.nscale.com/v1",
+    "openai": "https://api.openai.com/v1",
+    "gemini": "https://generativelanguage.googleapis.com/v1beta",
+    "anthropic": "https://api.anthropic.com/v1",
+    "cohere": "https://api.cohere.ai/v1",
+    "bedrock": "https://bedrock-runtime.us-east-1.amazonaws.com",
+    "openrouter": "https://openrouter.ai/api/v1",
+}
+
+def get_provider_endpoint(provider: str, model_name: str, incoming_path: str) -> Optional[str]:
+    """
+    Constructs the full provider endpoint URL based on the provider and incoming request path.
+    Supports both hardcoded providers and custom OpenAI-compatible providers via environment variables.
+    """
+    # First, check the hardcoded map
+    base_url = PROVIDER_URL_MAP.get(provider)
+
+    # If not found, check for custom provider via environment variable
+    if not base_url:
+        api_base_env = f"{provider.upper()}_API_BASE"
+        base_url = os.getenv(api_base_env)
+        if not base_url:
+            return None
+
+    # Determine the specific action from the incoming path (e.g., 'chat/completions')
+    action = incoming_path.split('/v1/', 1)[-1] if '/v1/' in incoming_path else incoming_path
+
+    # --- Provider-specific endpoint structures ---
+    if provider == "gemini":
+        if action == "chat/completions":
+            return f"{base_url}/models/{model_name}:generateContent"
+        elif action == "embeddings":
+            return f"{base_url}/models/{model_name}:embedContent"
+    
+    elif provider == "anthropic":
+        if action == "chat/completions":
+            return f"{base_url}/messages"
+
+    elif provider == "cohere":
+        if action == "chat/completions":
+            return f"{base_url}/chat"
+        elif action == "embeddings":
+            return f"{base_url}/embed"
+
+    # Default for OpenAI-compatible providers
+    # Most of these have /v1 in the base URL already, so we just append the action.
+    if base_url.endswith(("/v1", "/v1/openai")):
+        return f"{base_url}/{action}"
+    
+    # Fallback for other cases
+    return f"{base_url}/v1/{action}"

src/proxy_app/quota_viewer.py

@@ -0,0 +1,1596 @@
+# SPDX-License-Identifier: MIT
+# Copyright (c) 2026 Mirrowel
+
+"""
+Lightweight Quota Stats Viewer TUI.
+
+Connects to a running proxy to display quota and usage statistics.
+Uses only httpx + rich (no heavy rotator_library imports).
+
+TODO: Missing Features & Improvements
+======================================
+
+Display Improvements:
+- [ ] Add color legend/help screen explaining status colors and symbols
+- [ ] Show credential email/project ID if available (currently just filename)
+- [ ] Add keyboard shortcut hints (e.g., "Press ? for help")
+- [ ] Support terminal resize / responsive layout
+
+Global Stats Fix:
+- [ ] HACK: Global requests currently set to current period requests only
+      (see client.py get_quota_stats). This doesn't include archived stats.
+      Fix requires tracking archived requests per quota group in usage_manager.py
+      to avoid double-counting models that share quota groups.
+
+Data & Refresh:
+- [ ] Auto-refresh option (configurable interval)
+- [ ] Show last refresh timestamp more prominently
+- [ ] Cache invalidation when switching between current/global view
+- [ ] Support for non-OAuth providers (API keys like nvapi-*, gsk_*, etc.)
+
+Remote Management:
+- [ ] Test connection before saving remote
+- [ ] Import/export remote configurations
+- [ ] SSH tunnel support for remote proxies
+
+Quota Groups:
+- [ ] Show which models are in each quota group (expandable)
+- [ ] Historical quota usage graphs (if data available)
+- [ ] Alerts/notifications when quota is low
+
+Credential Details:
+- [ ] Show per-model breakdown within quota groups
+- [ ] Edit credential priority/tier manually
+- [ ] Disable/enable individual credentials
+"""
+
+import os
+import re
+import sys
+import time
+from datetime import datetime, timezone
+from typing import Any, Dict, List, Optional, Tuple
+
+import httpx
+from rich.console import Console
+from rich.panel import Panel
+from rich.progress import BarColumn, Progress, TextColumn
+from rich.prompt import Prompt
+from rich.table import Table
+from rich.text import Text
+
+from .quota_viewer_config import QuotaViewerConfig
+
+
+def clear_screen():
+    """Clear the terminal screen."""
+    os.system("cls" if os.name == "nt" else "clear")
+
+
+def format_tokens(count: int) -> str:
+    """Format token count for display (e.g., 125000 -> 125k)."""
+    if count >= 1_000_000:
+        return f"{count / 1_000_000:.1f}M"
+    elif count >= 1_000:
+        return f"{count / 1_000:.0f}k"
+    return str(count)
+
+
+def format_cost(cost: Optional[float]) -> str:
+    """Format cost for display."""
+    if cost is None or cost == 0:
+        return "-"
+    if cost < 0.01:
+        return f"${cost:.4f}"
+    return f"${cost:.2f}"
+
+
+def format_time_ago(timestamp: Optional[float]) -> str:
+    """Format timestamp as relative time (e.g., '5 min ago')."""
+    if not timestamp:
+        return "Never"
+    try:
+        delta = time.time() - timestamp
+        if delta < 60:
+            return f"{int(delta)}s ago"
+        elif delta < 3600:
+            return f"{int(delta / 60)} min ago"
+        elif delta < 86400:
+            return f"{int(delta / 3600)}h ago"
+        else:
+            return f"{int(delta / 86400)}d ago"
+    except (ValueError, OSError):
+        return "Unknown"
+
+
+def format_reset_time(iso_time: Optional[str]) -> str:
+    """Format ISO time string for display."""
+    if not iso_time:
+        return "-"
+    try:
+        dt = datetime.fromisoformat(iso_time.replace("Z", "+00:00"))
+        # Convert to local time
+        local_dt = dt.astimezone()
+        return local_dt.strftime("%b %d %H:%M")
+    except (ValueError, AttributeError):
+        return iso_time[:16] if iso_time else "-"
+
+
+def create_progress_bar(percent: Optional[int], width: int = 10) -> str:
+    """Create a text-based progress bar."""
+    if percent is None:
+        return "░" * width
+    filled = int(percent / 100 * width)
+    return "▓" * filled + "░" * (width - filled)
+
+
+def is_local_host(host: str) -> bool:
+    """Check if host is a local/private address (should use http, not https)."""
+    if host in ("localhost", "127.0.0.1", "::1", "0.0.0.0", "::"):
+        return True
+    # Private IP ranges
+    if host.startswith("192.168.") or host.startswith("10."):
+        return True
+    if host.startswith("172."):
+        # 172.16.0.0 - 172.31.255.255
+        try:
+            second_octet = int(host.split(".")[1])
+            if 16 <= second_octet <= 31:
+                return True
+        except (ValueError, IndexError):
+            pass
+    return False
+
+
+def normalize_host_for_connection(host: str) -> str:
+    """
+    Convert bind addresses to connectable addresses.
+
+    0.0.0.0 and :: are valid for binding a server to all interfaces,
+    but clients cannot connect to them. Translate to loopback addresses.
+    """
+    if host == "0.0.0.0":
+        return "127.0.0.1"
+    if host == "::":
+        return "::1"
+    return host
+
+
+def get_scheme_for_host(host: str, port: int) -> str:
+    """Determine http or https scheme based on host and port."""
+    if port == 443:
+        return "https"
+    if is_local_host(host):
+        return "http"
+    # For external domains, default to https
+    if "." in host:
+        return "https"
+    return "http"
+
+
+def is_full_url(host: str) -> bool:
+    """Check if host is already a full URL (starts with http:// or https://)."""
+    return host.startswith("http://") or host.startswith("https://")
+
+
+def format_cooldown(seconds: int) -> str:
+    """Format cooldown seconds as human-readable string."""
+    if seconds < 60:
+        return f"{seconds}s"
+    elif seconds < 3600:
+        mins = seconds // 60
+        secs = seconds % 60
+        return f"{mins}m {secs}s" if secs > 0 else f"{mins}m"
+    else:
+        hours = seconds // 3600
+        mins = (seconds % 3600) // 60
+        return f"{hours}h {mins}m" if mins > 0 else f"{hours}h"
+
+
+def natural_sort_key(item: Dict[str, Any]) -> List:
+    """
+    Generate a sort key for natural/numeric sorting.
+
+    Sorts credentials like proj-1, proj-2, proj-10 correctly
+    instead of alphabetically (proj-1, proj-10, proj-2).
+    """
+    identifier = item.get("identifier", "")
+    # Split into text and numeric parts
+    parts = re.split(r"(\d+)", identifier)
+    return [int(p) if p.isdigit() else p.lower() for p in parts]
+
+
+class QuotaViewer:
+    """Main Quota Viewer TUI class."""
+
+    def __init__(self, config: Optional[QuotaViewerConfig] = None):
+        """
+        Initialize the viewer.
+
+        Args:
+            config: Optional config object. If not provided, one will be created.
+        """
+        self.console = Console()
+        self.config = config or QuotaViewerConfig()
+        self.config.sync_with_launcher_config()
+
+        self.current_remote: Optional[Dict[str, Any]] = None
+        self.cached_stats: Optional[Dict[str, Any]] = None
+        self.last_error: Optional[str] = None
+        self.running = True
+        self.view_mode = "current"  # "current" or "global"
+
+    def _get_headers(self) -> Dict[str, str]:
+        """Get HTTP headers including auth if configured."""
+        headers = {}
+        if self.current_remote and self.current_remote.get("api_key"):
+            headers["Authorization"] = f"Bearer {self.current_remote['api_key']}"
+        return headers
+
+    def _get_base_url(self) -> str:
+        """Get base URL for the current remote."""
+        if not self.current_remote:
+            return "http://127.0.0.1:8000"
+        host = self.current_remote.get("host", "127.0.0.1")
+        host = normalize_host_for_connection(host)
+
+        # If host is a full URL, use it directly (strip trailing slash)
+        if is_full_url(host):
+            return host.rstrip("/")
+
+        # Otherwise construct from host:port
+        port = self.current_remote.get("port", 8000)
+        scheme = get_scheme_for_host(host, port)
+        return f"{scheme}://{host}:{port}"
+
+    def _build_endpoint_url(self, endpoint: str) -> str:
+        """
+        Build a full endpoint URL with smart path handling.
+
+        Handles cases where base URL already contains a path (e.g., /v1):
+        - Base: "https://api.example.com/v1", Endpoint: "/v1/quota-stats"
+          -> "https://api.example.com/v1/quota-stats" (no duplication)
+        - Base: "http://localhost:8000", Endpoint: "/v1/quota-stats"
+          -> "http://localhost:8000/v1/quota-stats"
+
+        Args:
+            endpoint: The endpoint path (e.g., "/v1/quota-stats")
+
+        Returns:
+            Full URL string
+        """
+        base_url = self._get_base_url()
+        endpoint = endpoint.lstrip("/")
+
+        # Check if base URL already ends with a path segment that matches
+        # the start of the endpoint (e.g., base ends with /v1, endpoint starts with v1/)
+        from urllib.parse import urlparse
+
+        parsed = urlparse(base_url)
+        base_path = parsed.path.rstrip("/")
+
+        # If base has a path and endpoint starts with the same segment, avoid duplication
+        if base_path:
+            # e.g., base_path = "/v1", endpoint = "v1/quota-stats"
+            # We want to produce "/v1/quota-stats", not "/v1/v1/quota-stats"
+            base_segments = base_path.split("/")
+            endpoint_segments = endpoint.split("/")
+
+            # Check if first endpoint segment matches last base segment
+            if base_segments and endpoint_segments:
+                if base_segments[-1] == endpoint_segments[0]:
+                    # Skip the duplicated segment in endpoint
+                    endpoint = "/".join(endpoint_segments[1:])
+
+        return f"{base_url}/{endpoint}"
+
+    def check_connection(
+        self, remote: Dict[str, Any], timeout: float = 3.0
+    ) -> Tuple[bool, str]:
+        """
+        Check if a remote proxy is reachable.
+
+        Args:
+            remote: Remote configuration dict
+            timeout: Connection timeout in seconds
+
+        Returns:
+            Tuple of (is_online, status_message)
+        """
+        host = remote.get("host", "127.0.0.1")
+        host = normalize_host_for_connection(host)
+
+        # If host is a full URL, extract scheme and netloc to hit root
+        if is_full_url(host):
+            from urllib.parse import urlparse
+
+            parsed = urlparse(host)
+            # Hit the root domain, not the path (e.g., /v1 would 404)
+            url = f"{parsed.scheme}://{parsed.netloc}/"
+        else:
+            port = remote.get("port", 8000)
+            scheme = get_scheme_for_host(host, port)
+            url = f"{scheme}://{host}:{port}/"
+
+        headers = {}
+        if remote.get("api_key"):
+            headers["Authorization"] = f"Bearer {remote['api_key']}"
+
+        try:
+            with httpx.Client(timeout=timeout) as client:
+                response = client.get(url, headers=headers)
+                if response.status_code == 200:
+                    return True, "Online"
+                elif response.status_code == 401:
+                    return False, "Auth failed"
+                else:
+                    return False, f"HTTP {response.status_code}"
+        except httpx.ConnectError:
+            return False, "Offline"
+        except httpx.TimeoutException:
+            return False, "Timeout"
+        except Exception as e:
+            return False, str(e)[:20]
+
+    def fetch_stats(self, provider: Optional[str] = None) -> Optional[Dict[str, Any]]:
+        """
+        Fetch quota stats from the current remote.
+
+        Args:
+            provider: Optional provider filter
+
+        Returns:
+            Stats dict or None on failure
+        """
+        url = self._build_endpoint_url("/v1/quota-stats")
+        if provider:
+            url += f"?provider={provider}"
+
+        try:
+            with httpx.Client(timeout=30.0) as client:
+                response = client.get(url, headers=self._get_headers())
+
+                if response.status_code == 401:
+                    self.last_error = "Authentication failed. Check API key."
+                    return None
+                elif response.status_code != 200:
+                    self.last_error = (
+                        f"HTTP {response.status_code}: {response.text[:100]}"
+                    )
+                    return None
+
+                self.cached_stats = response.json()
+                self.last_error = None
+                return self.cached_stats
+
+        except httpx.ConnectError:
+            self.last_error = "Connection failed. Is the proxy running?"
+            return None
+        except httpx.TimeoutException:
+            self.last_error = "Request timed out."
+            return None
+        except Exception as e:
+            self.last_error = str(e)
+            return None
+
+    def _merge_provider_stats(self, provider: str, result: Dict[str, Any]) -> None:
+        """
+        Merge provider-specific stats into the existing cache.
+
+        Updates just the specified provider's data and recalculates the
+        summary fields to reflect the change.
+
+        Args:
+            provider: Provider name that was refreshed
+            result: API response containing the refreshed provider data
+        """
+        if not self.cached_stats:
+            self.cached_stats = result
+            return
+
+        # Merge provider data
+        if "providers" in result and provider in result["providers"]:
+            if "providers" not in self.cached_stats:
+                self.cached_stats["providers"] = {}
+            self.cached_stats["providers"][provider] = result["providers"][provider]
+
+        # Update timestamp
+        if "timestamp" in result:
+            self.cached_stats["timestamp"] = result["timestamp"]
+
+        # Recalculate summary from all providers
+        self._recalculate_summary()
+
+    def _recalculate_summary(self) -> None:
+        """
+        Recalculate summary fields from all provider data in cache.
+
+        Updates both 'summary' and 'global_summary' based on current
+        provider stats.
+        """
+        providers = self.cached_stats.get("providers", {})
+        if not providers:
+            return
+
+        # Calculate summary from all providers
+        total_creds = 0
+        active_creds = 0
+        exhausted_creds = 0
+        total_requests = 0
+        total_input_cached = 0
+        total_input_uncached = 0
+        total_output = 0
+        total_cost = 0.0
+
+        for prov_stats in providers.values():
+            total_creds += prov_stats.get("credential_count", 0)
+            active_creds += prov_stats.get("active_count", 0)
+            exhausted_creds += prov_stats.get("exhausted_count", 0)
+            total_requests += prov_stats.get("total_requests", 0)
+
+            tokens = prov_stats.get("tokens", {})
+            total_input_cached += tokens.get("input_cached", 0)
+            total_input_uncached += tokens.get("input_uncached", 0)
+            total_output += tokens.get("output", 0)
+
+            cost = prov_stats.get("approx_cost")
+            if cost:
+                total_cost += cost
+
+        total_input = total_input_cached + total_input_uncached
+        input_cache_pct = (
+            round(total_input_cached / total_input * 100, 1) if total_input > 0 else 0
+        )
+
+        self.cached_stats["summary"] = {
+            "total_providers": len(providers),
+            "total_credentials": total_creds,
+            "active_credentials": active_creds,
+            "exhausted_credentials": exhausted_creds,
+            "total_requests": total_requests,
+            "tokens": {
+                "input_cached": total_input_cached,
+                "input_uncached": total_input_uncached,
+                "input_cache_pct": input_cache_pct,
+                "output": total_output,
+            },
+            "approx_total_cost": total_cost if total_cost > 0 else None,
+        }
+
+        # Also recalculate global_summary if it exists
+        if "global_summary" in self.cached_stats:
+            global_total_requests = 0
+            global_input_cached = 0
+            global_input_uncached = 0
+            global_output = 0
+            global_cost = 0.0
+
+            for prov_stats in providers.values():
+                global_data = prov_stats.get("global", prov_stats)
+                global_total_requests += global_data.get("total_requests", 0)
+
+                tokens = global_data.get("tokens", {})
+                global_input_cached += tokens.get("input_cached", 0)
+                global_input_uncached += tokens.get("input_uncached", 0)
+                global_output += tokens.get("output", 0)
+
+                cost = global_data.get("approx_cost")
+                if cost:
+                    global_cost += cost
+
+            global_total_input = global_input_cached + global_input_uncached
+            global_cache_pct = (
+                round(global_input_cached / global_total_input * 100, 1)
+                if global_total_input > 0
+                else 0
+            )
+
+            self.cached_stats["global_summary"] = {
+                "total_providers": len(providers),
+                "total_credentials": total_creds,
+                "total_requests": global_total_requests,
+                "tokens": {
+                    "input_cached": global_input_cached,
+                    "input_uncached": global_input_uncached,
+                    "input_cache_pct": global_cache_pct,
+                    "output": global_output,
+                },
+                "approx_total_cost": global_cost if global_cost > 0 else None,
+            }
+
+    def post_action(
+        self,
+        action: str,
+        scope: str = "all",
+        provider: Optional[str] = None,
+        credential: Optional[str] = None,
+    ) -> Optional[Dict[str, Any]]:
+        """
+        Post a refresh action to the proxy.
+
+        Args:
+            action: "reload" or "force_refresh"
+            scope: "all", "provider", or "credential"
+            provider: Provider name (required for scope != "all")
+            credential: Credential identifier (required for scope == "credential")
+
+        Returns:
+            Response dict or None on failure
+        """
+        url = self._build_endpoint_url("/v1/quota-stats")
+        payload = {
+            "action": action,
+            "scope": scope,
+        }
+        if provider:
+            payload["provider"] = provider
+        if credential:
+            payload["credential"] = credential
+
+        try:
+            with httpx.Client(timeout=60.0) as client:
+                response = client.post(url, headers=self._get_headers(), json=payload)
+
+                if response.status_code == 401:
+                    self.last_error = "Authentication failed. Check API key."
+                    return None
+                elif response.status_code != 200:
+                    self.last_error = (
+                        f"HTTP {response.status_code}: {response.text[:100]}"
+                    )
+                    return None
+
+                result = response.json()
+
+                # If scope is provider-specific, merge into existing cache
+                if scope == "provider" and provider and self.cached_stats:
+                    self._merge_provider_stats(provider, result)
+                else:
+                    # Full refresh - replace everything
+                    self.cached_stats = result
+
+                self.last_error = None
+                return result
+
+        except httpx.ConnectError:
+            self.last_error = "Connection failed. Is the proxy running?"
+            return None
+        except httpx.TimeoutException:
+            self.last_error = "Request timed out."
+            return None
+        except Exception as e:
+            self.last_error = str(e)
+            return None
+
+    # =========================================================================
+    # DISPLAY SCREENS
+    # =========================================================================
+
+    def show_connection_error(self) -> str:
+        """
+        Display connection error screen with options to configure remotes.
+
+        Returns:
+            User choice: 's' (switch), 'm' (manage), 'r' (retry), 'b' (back/exit)
+        """
+        clear_screen()
+
+        remote_name = (
+            self.current_remote.get("name", "Unknown")
+            if self.current_remote
+            else "None"
+        )
+        remote_host = self.current_remote.get("host", "") if self.current_remote else ""
+        remote_port = self.current_remote.get("port", "") if self.current_remote else ""
+
+        # Format connection display - handle full URLs
+        if is_full_url(remote_host):
+            connection_display = remote_host
+        elif remote_port:
+            connection_display = f"{remote_host}:{remote_port}"
+        else:
+            connection_display = remote_host
+
+        self.console.print(
+            Panel(
+                Text.from_markup(
+                    "[bold red]Connection Error[/bold red]\n\n"
+                    f"Remote: [bold]{remote_name}[/bold] ({connection_display})\n"
+                    f"Error: {self.last_error or 'Unknown error'}\n\n"
+                    "[bold]This tool requires the proxy to be running.[/bold]\n"
+                    "Start the proxy first, or configure a different remote.\n\n"
+                    "[dim]Tip: Select option 1 from the main menu to run the proxy.[/dim]"
+                ),
+                border_style="red",
+                expand=False,
+            )
+        )
+
+        self.console.print()
+        self.console.print("━" * 78)
+        self.console.print()
+        self.console.print("   S. Switch to a different remote")
+        self.console.print("   M. Manage remotes (add/edit/delete)")
+        self.console.print("   R. Retry connection")
+        self.console.print("   B. Back to main menu")
+        self.console.print()
+        self.console.print("━" * 78)
+
+        choice = Prompt.ask("Select option", default="B").strip().lower()
+
+        if choice in ("s", "m", "r", "b"):
+            return choice
+        return "b"  # Default to back for invalid input
+
+    def show_summary_screen(self):
+        """Display the main summary screen with all providers."""
+        clear_screen()
+
+        # Header
+        remote_name = (
+            self.current_remote.get("name", "Unknown")
+            if self.current_remote
+            else "None"
+        )
+        remote_host = self.current_remote.get("host", "") if self.current_remote else ""
+        remote_port = self.current_remote.get("port", "") if self.current_remote else ""
+
+        # Format connection display - handle full URLs
+        if is_full_url(remote_host):
+            connection_display = remote_host
+        elif remote_port:
+            connection_display = f"{remote_host}:{remote_port}"
+        else:
+            connection_display = remote_host
+
+        # Calculate data age
+        data_age = ""
+        if self.cached_stats and self.cached_stats.get("timestamp"):
+            age_seconds = int(time.time() - self.cached_stats["timestamp"])
+            data_age = f"Data age: {age_seconds}s"
+
+        # View mode indicator
+        if self.view_mode == "global":
+            view_label = "[magenta]📊 Global/Lifetime[/magenta]"
+        else:
+            view_label = "[cyan]📈 Current Period[/cyan]"
+
+        self.console.print("━" * 78)
+        self.console.print(
+            f"[bold cyan]📈 Quota & Usage Statistics[/bold cyan]  |  {view_label}"
+        )
+        self.console.print("━" * 78)
+        self.console.print(
+            f"Connected to: [bold]{remote_name}[/bold] ({connection_display}) "
+            f"[green]✅[/green] | {data_age}"
+        )
+        self.console.print()
+
+        if not self.cached_stats:
+            self.console.print("[yellow]No data available. Press R to reload.[/yellow]")
+        else:
+            # Build provider table
+            table = Table(
+                box=None, show_header=True, header_style="bold", padding=(0, 1)
+            )
+            table.add_column("Provider", style="cyan", min_width=10)
+            table.add_column("Creds", justify="center", min_width=5)
+            table.add_column("Quota Status", min_width=28)
+            table.add_column("Requests", justify="right", min_width=8)
+            table.add_column("Tokens (in/out)", min_width=20)
+            table.add_column("Cost", justify="right", min_width=6)
+
+            providers = self.cached_stats.get("providers", {})
+            provider_list = list(providers.keys())
+
+            for idx, (provider, prov_stats) in enumerate(providers.items(), 1):
+                cred_count = prov_stats.get("credential_count", 0)
+
+                # Use global stats if in global mode
+                if self.view_mode == "global":
+                    stats_source = prov_stats.get("global", prov_stats)
+                    total_requests = stats_source.get("total_requests", 0)
+                    tokens = stats_source.get("tokens", {})
+                    cost_value = stats_source.get("approx_cost")
+                else:
+                    total_requests = prov_stats.get("total_requests", 0)
+                    tokens = prov_stats.get("tokens", {})
+                    cost_value = prov_stats.get("approx_cost")
+
+                # Format tokens
+                input_total = tokens.get("input_cached", 0) + tokens.get(
+                    "input_uncached", 0
+                )
+                output = tokens.get("output", 0)
+                cache_pct = tokens.get("input_cache_pct", 0)
+                token_str = f"{format_tokens(input_total)}/{format_tokens(output)} ({cache_pct}% cached)"
+
+                # Format cost
+                cost_str = format_cost(cost_value)
+
+                # Build quota status string (for providers with quota groups)
+                quota_groups = prov_stats.get("quota_groups", {})
+                if quota_groups:
+                    quota_lines = []
+                    for group_name, group_stats in quota_groups.items():
+                        # Use remaining requests (not used) so percentage matches displayed value
+                        total_remaining = group_stats.get("total_requests_remaining", 0)
+                        total_max = group_stats.get("total_requests_max", 0)
+                        total_pct = group_stats.get("total_remaining_pct")
+                        tiers = group_stats.get("tiers", {})
+
+                        # Format tier info: "5(15)f/2s" = 5 active out of 15 free, 2 standard all active
+                        # Sort by priority (lower number = higher priority, appears first)
+                        tier_parts = []
+                        sorted_tiers = sorted(
+                            tiers.items(), key=lambda x: x[1].get("priority", 10)
+                        )
+                        for tier_name, tier_info in sorted_tiers:
+                            if tier_name == "unknown":
+                                continue  # Skip unknown tiers in display
+                            total_t = tier_info.get("total", 0)
+                            active_t = tier_info.get("active", 0)
+                            # Use first letter: standard-tier -> s, free-tier -> f
+                            short = tier_name.replace("-tier", "")[0]
+
+                            if active_t < total_t:
+                                # Some exhausted - show active(total)
+                                tier_parts.append(f"{active_t}({total_t}){short}")
+                            else:
+                                # All active - just show total
+                                tier_parts.append(f"{total_t}{short}")
+                        tier_str = "/".join(tier_parts) if tier_parts else ""
+
+                        # Determine color based purely on remaining percentage
+                        if total_pct is not None:
+                            if total_pct <= 10:
+                                color = "red"
+                            elif total_pct < 30:
+                                color = "yellow"
+                            else:
+                                color = "green"
+                        else:
+                            color = "dim"
+
+                        bar = create_progress_bar(total_pct)
+                        pct_str = f"{total_pct}%" if total_pct is not None else "?"
+
+                        # Build status suffix (just tiers now, no outer parens)
+                        status = tier_str
+
+                        # Fixed-width format for aligned bars
+                        # Adjust these to change column spacing:
+                        QUOTA_NAME_WIDTH = 10  # name + colon, left-aligned
+                        QUOTA_USAGE_WIDTH = (
+                            12  # remaining/max ratio, right-aligned (handles 100k+)
+                        )
+                        display_name = group_name[: QUOTA_NAME_WIDTH - 1]
+                        usage_str = f"{total_remaining}/{total_max}"
+                        quota_lines.append(
+                            f"[{color}]{display_name + ':':<{QUOTA_NAME_WIDTH}}{usage_str:>{QUOTA_USAGE_WIDTH}} {pct_str:>4} {bar}[/{color}] {status}"
+                        )
+
+                    # First line goes in the main row
+                    first_quota = quota_lines[0] if quota_lines else "-"
+                    table.add_row(
+                        provider,
+                        str(cred_count),
+                        first_quota,
+                        str(total_requests),
+                        token_str,
+                        cost_str,
+                    )
+                    # Additional quota lines as sub-rows
+                    for quota_line in quota_lines[1:]:
+                        table.add_row("", "", quota_line, "", "", "")
+                else:
+                    # No quota groups
+                    table.add_row(
+                        provider,
+                        str(cred_count),
+                        "-",
+                        str(total_requests),
+                        token_str,
+                        cost_str,
+                    )
+
+                # Add separator between providers (except last)
+                if idx < len(providers):
+                    table.add_row(
+                        "─" * 10, "─" * 4, "─" * 26, "─" * 7, "─" * 20, "─" * 6
+                    )
+
+            self.console.print(table)
+
+            # Summary line - use global_summary if in global mode
+            if self.view_mode == "global":
+                summary = self.cached_stats.get(
+                    "global_summary", self.cached_stats.get("summary", {})
+                )
+            else:
+                summary = self.cached_stats.get("summary", {})
+
+            total_creds = summary.get("total_credentials", 0)
+            total_requests = summary.get("total_requests", 0)
+            total_tokens = summary.get("tokens", {})
+            total_input = total_tokens.get("input_cached", 0) + total_tokens.get(
+                "input_uncached", 0
+            )
+            total_output = total_tokens.get("output", 0)
+            total_cost = format_cost(summary.get("approx_total_cost"))
+
+            self.console.print()
+            self.console.print(
+                f"[bold]Total:[/bold] {total_creds} credentials | "
+                f"{total_requests} requests | "
+                f"{format_tokens(total_input)}/{format_tokens(total_output)} tokens | "
+                f"{total_cost} cost"
+            )
+
+        # Menu
+        self.console.print()
+        self.console.print("━" * 78)
+        self.console.print()
+
+        # Build provider menu options
+        providers = self.cached_stats.get("providers", {}) if self.cached_stats else {}
+        provider_list = list(providers.keys())
+
+        for idx, provider in enumerate(provider_list, 1):
+            self.console.print(f"   {idx}. View [cyan]{provider}[/cyan] details")
+
+        self.console.print()
+        self.console.print("   G. Toggle view mode (current/global)")
+        self.console.print("   R. Reload all stats (re-read from proxy)")
+        self.console.print("   S. Switch remote")
+        self.console.print("   M. Manage remotes")
+        self.console.print("   B. Back to main menu")
+        self.console.print()
+        self.console.print("━" * 78)
+
+        # Get input
+        valid_choices = [str(i) for i in range(1, len(provider_list) + 1)]
+        valid_choices.extend(["r", "R", "s", "S", "m", "M", "b", "B", "g", "G"])
+
+        choice = Prompt.ask("Select option", default="").strip()
+
+        if choice.lower() == "b":
+            self.running = False
+        elif choice == "":
+            # Empty input - just refresh the screen
+            pass
+        elif choice.lower() == "g":
+            # Toggle view mode
+            self.view_mode = "global" if self.view_mode == "current" else "current"
+        elif choice.lower() == "r":
+            with self.console.status("[bold]Reloading stats...", spinner="dots"):
+                self.post_action("reload", scope="all")
+        elif choice.lower() == "s":
+            self.show_switch_remote_screen()
+        elif choice.lower() == "m":
+            self.show_manage_remotes_screen()
+        elif choice.isdigit() and 1 <= int(choice) <= len(provider_list):
+            provider = provider_list[int(choice) - 1]
+            self.show_provider_detail_screen(provider)
+
+    def show_provider_detail_screen(self, provider: str):
+        """Display detailed stats for a specific provider."""
+        while True:
+            clear_screen()
+
+            # View mode indicator
+            if self.view_mode == "global":
+                view_label = "[magenta]Global/Lifetime[/magenta]"
+            else:
+                view_label = "[cyan]Current Period[/cyan]"
+
+            self.console.print("━" * 78)
+            self.console.print(
+                f"[bold cyan]📊 {provider.title()} - Detailed Stats[/bold cyan]  |  {view_label}"
+            )
+            self.console.print("━" * 78)
+            self.console.print()
+
+            if not self.cached_stats:
+                self.console.print("[yellow]No data available.[/yellow]")
+            else:
+                prov_stats = self.cached_stats.get("providers", {}).get(provider, {})
+                credentials = prov_stats.get("credentials", [])
+
+                # Sort credentials naturally (1, 2, 10 not 1, 10, 2)
+                credentials = sorted(credentials, key=natural_sort_key)
+
+                if not credentials:
+                    self.console.print(
+                        "[dim]No credentials configured for this provider.[/dim]"
+                    )
+                else:
+                    for idx, cred in enumerate(credentials, 1):
+                        self._render_credential_panel(idx, cred, provider)
+                        self.console.print()
+
+            # Menu
+            self.console.print("━" * 78)
+            self.console.print()
+            self.console.print("   G.  Toggle view mode (current/global)")
+            self.console.print("   R.  Reload stats (from proxy cache)")
+            self.console.print("   RA. Reload all stats")
+
+            # Force refresh options (only for providers that support it)
+            has_quota_groups = bool(
+                self.cached_stats
+                and self.cached_stats.get("providers", {})
+                .get(provider, {})
+                .get("quota_groups")
+            )
+
+            if has_quota_groups:
+                self.console.print()
+                self.console.print(
+                    f"   F.  [yellow]Force refresh ALL {provider} quotas from API[/yellow]"
+                )
+                credentials = (
+                    self.cached_stats.get("providers", {})
+                    .get(provider, {})
+                    .get("credentials", [])
+                    if self.cached_stats
+                    else []
+                )
+                # Sort credentials naturally
+                credentials = sorted(credentials, key=natural_sort_key)
+                for idx, cred in enumerate(credentials, 1):
+                    identifier = cred.get("identifier", f"credential {idx}")
+                    email = cred.get("email", identifier)
+                    self.console.print(
+                        f"   F{idx}. Force refresh [{idx}] only ({email})"
+                    )
+
+            self.console.print()
+            self.console.print("   B.  Back to summary")
+            self.console.print()
+            self.console.print("━" * 78)
+
+            choice = Prompt.ask("Select option", default="B").strip().upper()
+
+            if choice == "B":
+                break
+            elif choice == "G":
+                # Toggle view mode
+                self.view_mode = "global" if self.view_mode == "current" else "current"
+            elif choice == "R":
+                with self.console.status(
+                    f"[bold]Reloading {provider} stats...", spinner="dots"
+                ):
+                    self.post_action("reload", scope="provider", provider=provider)
+            elif choice == "RA":
+                with self.console.status(
+                    "[bold]Reloading all stats...", spinner="dots"
+                ):
+                    self.post_action("reload", scope="all")
+            elif choice == "F" and has_quota_groups:
+                result = None
+                with self.console.status(
+                    f"[bold]Fetching live quota for ALL {provider} credentials...",
+                    spinner="dots",
+                ):
+                    result = self.post_action(
+                        "force_refresh", scope="provider", provider=provider
+                    )
+                # Handle result OUTSIDE spinner
+                if result and result.get("refresh_result"):
+                    rr = result["refresh_result"]
+                    self.console.print(
+                        f"\n[green]Refreshed {rr.get('credentials_refreshed', 0)} credentials "
+                        f"in {rr.get('duration_ms', 0)}ms[/green]"
+                    )
+                    if rr.get("errors"):
+                        for err in rr["errors"]:
+                            self.console.print(f"[red]  Error: {err}[/red]")
+                    Prompt.ask("Press Enter to continue", default="")
+            elif choice.startswith("F") and choice[1:].isdigit() and has_quota_groups:
+                idx = int(choice[1:])
+                credentials = (
+                    self.cached_stats.get("providers", {})
+                    .get(provider, {})
+                    .get("credentials", [])
+                    if self.cached_stats
+                    else []
+                )
+                # Sort credentials naturally to match display order
+                credentials = sorted(credentials, key=natural_sort_key)
+                if 1 <= idx <= len(credentials):
+                    cred = credentials[idx - 1]
+                    cred_id = cred.get("identifier", "")
+                    email = cred.get("email", cred_id)
+                    result = None
+                    with self.console.status(
+                        f"[bold]Fetching live quota for {email}...", spinner="dots"
+                    ):
+                        result = self.post_action(
+                            "force_refresh",
+                            scope="credential",
+                            provider=provider,
+                            credential=cred_id,
+                        )
+                    # Handle result OUTSIDE spinner
+                    if result and result.get("refresh_result"):
+                        rr = result["refresh_result"]
+                        self.console.print(
+                            f"\n[green]Refreshed in {rr.get('duration_ms', 0)}ms[/green]"
+                        )
+                        if rr.get("errors"):
+                            for err in rr["errors"]:
+                                self.console.print(f"[red]  Error: {err}[/red]")
+                        Prompt.ask("Press Enter to continue", default="")
+
+    def _render_credential_panel(self, idx: int, cred: Dict[str, Any], provider: str):
+        """Render a single credential as a panel."""
+        identifier = cred.get("identifier", f"credential {idx}")
+        email = cred.get("email")
+        tier = cred.get("tier", "")
+        status = cred.get("status", "unknown")
+
+        # Check for active cooldowns
+        key_cooldown = cred.get("key_cooldown_remaining")
+        model_cooldowns = cred.get("model_cooldowns", {})
+        has_cooldown = key_cooldown or model_cooldowns
+
+        # Status indicator
+        if status == "exhausted":
+            status_icon = "[red]⛔ Exhausted[/red]"
+        elif status == "cooldown" or has_cooldown:
+            if key_cooldown:
+                status_icon = f"[yellow]⚠️ Cooldown ({format_cooldown(int(key_cooldown))})[/yellow]"
+            else:
+                status_icon = "[yellow]⚠️ Cooldown[/yellow]"
+        else:
+            status_icon = "[green]✅ Active[/green]"
+
+        # Header line
+        display_name = email if email else identifier
+        tier_str = f" ({tier})" if tier else ""
+        header = f"[{idx}] {display_name}{tier_str} {status_icon}"
+
+        # Use global stats if in global mode
+        if self.view_mode == "global":
+            stats_source = cred.get("global", cred)
+        else:
+            stats_source = cred
+
+        # Stats line
+        last_used = format_time_ago(cred.get("last_used_ts"))  # Always from current
+        requests = stats_source.get("requests", 0)
+        tokens = stats_source.get("tokens", {})
+        input_total = tokens.get("input_cached", 0) + tokens.get("input_uncached", 0)
+        output = tokens.get("output", 0)
+        cost = format_cost(stats_source.get("approx_cost"))
+
+        stats_line = (
+            f"Last used: {last_used} | Requests: {requests} | "
+            f"Tokens: {format_tokens(input_total)}/{format_tokens(output)}"
+        )
+        if cost != "-":
+            stats_line += f" | Cost: {cost}"
+
+        # Build panel content
+        content_lines = [
+            f"[dim]{stats_line}[/dim]",
+        ]
+
+        # Model groups (for providers with quota tracking)
+        model_groups = cred.get("model_groups", {})
+
+        # Show cooldowns grouped by quota group (if model_groups exist)
+        if model_cooldowns:
+            if model_groups:
+                # Group cooldowns by quota group
+                group_cooldowns: Dict[
+                    str, int
+                ] = {}  # group_name -> max_remaining_seconds
+                ungrouped_cooldowns: List[Tuple[str, int]] = []
+
+                for model_name, cooldown_info in model_cooldowns.items():
+                    remaining = cooldown_info.get("remaining_seconds", 0)
+                    if remaining <= 0:
+                        continue
+
+                    # Find which group this model belongs to
+                    clean_model = model_name.split("/")[-1]
+                    found_group = None
+                    for group_name, group_info in model_groups.items():
+                        group_models = group_info.get("models", [])
+                        if clean_model in group_models:
+                            found_group = group_name
+                            break
+
+                    if found_group:
+                        group_cooldowns[found_group] = max(
+                            group_cooldowns.get(found_group, 0), remaining
+                        )
+                    else:
+                        ungrouped_cooldowns.append((model_name, remaining))
+
+                if group_cooldowns or ungrouped_cooldowns:
+                    content_lines.append("")
+                    content_lines.append("[yellow]Active Cooldowns:[/yellow]")
+
+                    # Show grouped cooldowns
+                    for group_name in sorted(group_cooldowns.keys()):
+                        remaining = group_cooldowns[group_name]
+                        content_lines.append(
+                            f"  [yellow]⏱️ {group_name}: {format_cooldown(remaining)}[/yellow]"
+                        )
+
+                    # Show ungrouped (shouldn't happen often)
+                    for model_name, remaining in ungrouped_cooldowns:
+                        short_model = model_name.split("/")[-1][:35]
+                        content_lines.append(
+                            f"  [yellow]⏱️ {short_model}: {format_cooldown(remaining)}[/yellow]"
+                        )
+            else:
+                # No model groups - show per-model cooldowns
+                content_lines.append("")
+                content_lines.append("[yellow]Active Cooldowns:[/yellow]")
+                for model_name, cooldown_info in model_cooldowns.items():
+                    remaining = cooldown_info.get("remaining_seconds", 0)
+                    if remaining > 0:
+                        short_model = model_name.split("/")[-1][:35]
+                        content_lines.append(
+                            f"  [yellow]⏱��� {short_model}: {format_cooldown(int(remaining))}[/yellow]"
+                        )
+
+        # Display model groups with quota info
+        if model_groups:
+            content_lines.append("")
+            for group_name, group_stats in model_groups.items():
+                remaining_pct = group_stats.get("remaining_pct")
+                requests_used = group_stats.get("requests_used", 0)
+                requests_max = group_stats.get("requests_max")
+                requests_remaining = group_stats.get("requests_remaining")
+                is_exhausted = group_stats.get("is_exhausted", False)
+                reset_time = format_reset_time(group_stats.get("reset_time_iso"))
+                confidence = group_stats.get("confidence", "low")
+
+                # Format display - use requests_remaining/max format
+                if requests_remaining is None and requests_max:
+                    requests_remaining = max(0, requests_max - requests_used)
+                display = group_stats.get(
+                    "display", f"{requests_remaining or 0}/{requests_max or '?'}"
+                )
+                bar = create_progress_bar(remaining_pct)
+
+                # Build status text - always show reset time if available
+                has_reset_time = reset_time and reset_time != "-"
+
+                # Color based on status
+                if is_exhausted:
+                    color = "red"
+                    if has_reset_time:
+                        status_text = f"⛔ Resets: {reset_time}"
+                    else:
+                        status_text = "⛔ EXHAUSTED"
+                elif remaining_pct is not None and remaining_pct < 20:
+                    color = "yellow"
+                    if has_reset_time:
+                        status_text = f"⚠️ Resets: {reset_time}"
+                    else:
+                        status_text = "⚠️ LOW"
+                else:
+                    color = "green"
+                    if has_reset_time:
+                        status_text = f"Resets: {reset_time}"
+                    else:
+                        status_text = ""  # Hide if unused/no reset time
+
+                # Confidence indicator
+                conf_indicator = ""
+                if confidence == "low":
+                    conf_indicator = " [dim](~)[/dim]"
+                elif confidence == "medium":
+                    conf_indicator = " [dim](?)[/dim]"
+
+                pct_str = f"{remaining_pct}%" if remaining_pct is not None else "?%"
+                content_lines.append(
+                    f"  [{color}]{group_name:<18} {display:<10} {pct_str:>4} {bar}[/{color}]  {status_text}{conf_indicator}"
+                )
+        else:
+            # For providers without quota groups, show model breakdown if available
+            models = cred.get("models", {})
+            if models:
+                content_lines.append("")
+                content_lines.append("  [dim]Models used:[/dim]")
+                for model_name, model_stats in models.items():
+                    req_count = model_stats.get("success_count", 0)
+                    model_cost = format_cost(model_stats.get("approx_cost"))
+                    # Shorten model name for display
+                    short_name = model_name.split("/")[-1][:30]
+                    content_lines.append(
+                        f"    {short_name}: {req_count} requests, {model_cost}"
+                    )
+
+        self.console.print(
+            Panel(
+                "\n".join(content_lines),
+                title=header,
+                title_align="left",
+                border_style="dim",
+                expand=True,
+            )
+        )
+
+    def show_switch_remote_screen(self):
+        """Display remote selection screen."""
+        clear_screen()
+
+        self.console.print("━" * 78)
+        self.console.print("[bold cyan]🔄 Switch Remote[/bold cyan]")
+        self.console.print("━" * 78)
+        self.console.print()
+
+        current_name = self.current_remote.get("name") if self.current_remote else None
+        self.console.print(f"Current: [bold]{current_name}[/bold]")
+        self.console.print()
+        self.console.print("Available remotes:")
+
+        remotes = self.config.get_remotes()
+        remote_status: List[Tuple[Dict, bool, str]] = []
+
+        # Check status of all remotes
+        with self.console.status("[dim]Checking remote status...", spinner="dots"):
+            for remote in remotes:
+                is_online, status_msg = self.check_connection(remote)
+                remote_status.append((remote, is_online, status_msg))
+
+        for idx, (remote, is_online, status_msg) in enumerate(remote_status, 1):
+            name = remote.get("name", "Unknown")
+            host = remote.get("host", "")
+            port = remote.get("port", "")
+
+            # Format connection display - handle full URLs
+            if is_full_url(host):
+                connection_display = host
+            elif port:
+                connection_display = f"{host}:{port}"
+            else:
+                connection_display = host
+
+            is_current = name == current_name
+            current_marker = " (current)" if is_current else ""
+
+            if is_online:
+                status_icon = "[green]✅ Online[/green]"
+            else:
+                status_icon = f"[red]⚠️ {status_msg}[/red]"
+
+            self.console.print(
+                f"   {idx}. {name:<20} {connection_display:<30} {status_icon}{current_marker}"
+            )
+
+        self.console.print()
+        self.console.print("━" * 78)
+        self.console.print()
+
+        choice = Prompt.ask(
+            f"Select remote (1-{len(remotes)}) or B to go back", default="B"
+        ).strip()
+
+        if choice.lower() == "b":
+            return
+
+        if choice.isdigit() and 1 <= int(choice) <= len(remotes):
+            selected = remotes[int(choice) - 1]
+            self.current_remote = selected
+            self.config.set_last_used(selected["name"])
+            self.cached_stats = None  # Clear cache
+
+            # Try to fetch stats from new remote
+            with self.console.status("[bold]Connecting...", spinner="dots"):
+                stats = self.fetch_stats()
+                if stats is None:
+                    # Try with API key from .env for Local
+                    if selected["name"] == "Local" and not selected.get("api_key"):
+                        env_key = self.config.get_api_key_from_env()
+                        if env_key:
+                            self.current_remote["api_key"] = env_key
+                            stats = self.fetch_stats()
+
+            if stats is None:
+                self.show_api_key_prompt()
+
+    def show_api_key_prompt(self):
+        """Prompt for API key when authentication fails."""
+        self.console.print()
+        self.console.print(
+            "[yellow]Authentication required or connection failed.[/yellow]"
+        )
+        self.console.print(f"Error: {self.last_error}")
+        self.console.print()
+
+        api_key = Prompt.ask(
+            "Enter API key (or press Enter to cancel)", default=""
+        ).strip()
+
+        if api_key:
+            self.current_remote["api_key"] = api_key
+            # Update config with new API key
+            self.config.update_remote(self.current_remote["name"], api_key=api_key)
+
+            # Try again
+            with self.console.status("[bold]Reconnecting...", spinner="dots"):
+                if self.fetch_stats() is None:
+                    self.console.print(f"[red]Still failed: {self.last_error}[/red]")
+                    Prompt.ask("Press Enter to continue", default="")
+        else:
+            self.console.print("[dim]Cancelled.[/dim]")
+            Prompt.ask("Press Enter to continue", default="")
+
+    def show_manage_remotes_screen(self):
+        """Display remote management screen."""
+        while True:
+            clear_screen()
+
+            self.console.print("━" * 78)
+            self.console.print("[bold cyan]⚙️ Manage Remotes[/bold cyan]")
+            self.console.print("━" * 78)
+            self.console.print()
+
+            remotes = self.config.get_remotes()
+
+            table = Table(box=None, show_header=True, header_style="bold")
+            table.add_column("#", style="dim", width=3)
+            table.add_column("Name", min_width=16)
+            table.add_column("Host", min_width=24)
+            table.add_column("Port", justify="right", width=6)
+            table.add_column("Default", width=8)
+
+            for idx, remote in enumerate(remotes, 1):
+                is_default = "★" if remote.get("is_default") else ""
+                table.add_row(
+                    str(idx),
+                    remote.get("name", ""),
+                    remote.get("host", ""),
+                    str(remote.get("port", 8000)),
+                    is_default,
+                )
+
+            self.console.print(table)
+
+            self.console.print()
+            self.console.print("━" * 78)
+            self.console.print()
+            self.console.print("   A. Add new remote")
+            self.console.print("   E. Edit remote (enter number, e.g., E1)")
+            self.console.print("   D. Delete remote (enter number, e.g., D1)")
+            self.console.print("   S. Set default remote")
+            self.console.print("   B. Back")
+            self.console.print()
+            self.console.print("━" * 78)
+
+            choice = Prompt.ask("Select option", default="B").strip().upper()
+
+            if choice == "B":
+                break
+            elif choice == "A":
+                self._add_remote_dialog()
+            elif choice == "S":
+                self._set_default_dialog(remotes)
+            elif choice.startswith("E") and choice[1:].isdigit():
+                idx = int(choice[1:])
+                if 1 <= idx <= len(remotes):
+                    self._edit_remote_dialog(remotes[idx - 1])
+            elif choice.startswith("D") and choice[1:].isdigit():
+                idx = int(choice[1:])
+                if 1 <= idx <= len(remotes):
+                    self._delete_remote_dialog(remotes[idx - 1])
+
+    def _add_remote_dialog(self):
+        """Dialog to add a new remote."""
+        self.console.print()
+        self.console.print("[bold]Add New Remote[/bold]")
+        self.console.print(
+            "[dim]For full URLs (e.g., https://api.example.com/v1), leave port empty[/dim]"
+        )
+        self.console.print()
+
+        name = Prompt.ask("Name", default="").strip()
+        if not name:
+            self.console.print("[dim]Cancelled.[/dim]")
+            return
+
+        host = Prompt.ask("Host (or full URL)", default="").strip()
+        if not host:
+            self.console.print("[dim]Cancelled.[/dim]")
+            return
+
+        # For full URLs, default to empty port
+        if is_full_url(host):
+            port_default = ""
+        else:
+            port_default = "8000"
+
+        port_str = Prompt.ask(
+            "Port (empty for full URLs)", default=port_default
+        ).strip()
+        if port_str == "":
+            port = ""
+        else:
+            try:
+                port = int(port_str)
+            except ValueError:
+                port = 8000
+
+        api_key = Prompt.ask("API Key (optional)", default="").strip() or None
+
+        if self.config.add_remote(name, host, port, api_key):
+            self.console.print(f"[green]Added remote '{name}'.[/green]")
+        else:
+            self.console.print(f"[red]Remote '{name}' already exists.[/red]")
+
+        Prompt.ask("Press Enter to continue", default="")
+
+    def _edit_remote_dialog(self, remote: Dict[str, Any]):
+        """Dialog to edit an existing remote."""
+        self.console.print()
+        self.console.print(f"[bold]Edit Remote: {remote['name']}[/bold]")
+        self.console.print(
+            "[dim]Press Enter to keep current value. For full URLs, leave port empty.[/dim]"
+        )
+        self.console.print()
+
+        new_name = Prompt.ask("Name", default=remote["name"]).strip()
+        new_host = Prompt.ask(
+            "Host (or full URL)", default=remote.get("host", "")
+        ).strip()
+
+        # Get current port, handle empty string
+        current_port = remote.get("port", "")
+        port_default = str(current_port) if current_port != "" else ""
+
+        new_port_str = Prompt.ask(
+            "Port (empty for full URLs)", default=port_default
+        ).strip()
+        if new_port_str == "":
+            new_port = ""
+        else:
+            try:
+                new_port = int(new_port_str)
+            except ValueError:
+                new_port = current_port if current_port != "" else 8000
+
+        current_key = remote.get("api_key", "") or ""
+        display_key = f"{current_key[:8]}..." if len(current_key) > 8 else current_key
+        new_key = Prompt.ask(
+            f"API Key (current: {display_key or 'none'})", default=""
+        ).strip()
+
+        updates = {}
+        if new_name != remote["name"]:
+            updates["new_name"] = new_name
+        if new_host != remote.get("host"):
+            updates["host"] = new_host
+        if new_port != remote.get("port"):
+            updates["port"] = new_port
+        if new_key:
+            updates["api_key"] = new_key
+
+        if updates:
+            if self.config.update_remote(remote["name"], **updates):
+                self.console.print("[green]Remote updated.[/green]")
+                # Update current_remote if it was the one being edited
+                if (
+                    self.current_remote
+                    and self.current_remote["name"] == remote["name"]
+                ):
+                    self.current_remote.update(updates)
+                    if "new_name" in updates:
+                        self.current_remote["name"] = updates["new_name"]
+            else:
+                self.console.print("[red]Failed to update remote.[/red]")
+        else:
+            self.console.print("[dim]No changes made.[/dim]")
+
+        Prompt.ask("Press Enter to continue", default="")
+
+    def _delete_remote_dialog(self, remote: Dict[str, Any]):
+        """Dialog to delete a remote."""
+        self.console.print()
+        self.console.print(f"[yellow]Delete remote '{remote['name']}'?[/yellow]")
+
+        confirm = Prompt.ask("Type 'yes' to confirm", default="no").strip().lower()
+
+        if confirm == "yes":
+            if self.config.delete_remote(remote["name"]):
+                self.console.print(f"[green]Deleted remote '{remote['name']}'.[/green]")
+                # If deleted current remote, switch to another
+                if (
+                    self.current_remote
+                    and self.current_remote["name"] == remote["name"]
+                ):
+                    self.current_remote = self.config.get_default_remote()
+                    self.cached_stats = None
+            else:
+                self.console.print(
+                    "[red]Cannot delete. At least one remote must exist.[/red]"
+                )
+        else:
+            self.console.print("[dim]Cancelled.[/dim]")
+
+        Prompt.ask("Press Enter to continue", default="")
+
+    def _set_default_dialog(self, remotes: List[Dict[str, Any]]):
+        """Dialog to set the default remote."""
+        self.console.print()
+        choice = Prompt.ask(f"Set default (1-{len(remotes)})", default="").strip()
+
+        if choice.isdigit() and 1 <= int(choice) <= len(remotes):
+            remote = remotes[int(choice) - 1]
+            if self.config.set_default_remote(remote["name"]):
+                self.console.print(
+                    f"[green]'{remote['name']}' is now the default.[/green]"
+                )
+            else:
+                self.console.print("[red]Failed to set default.[/red]")
+            Prompt.ask("Press Enter to continue", default="")
+
+    # =========================================================================
+    # MAIN LOOP
+    # =========================================================================
+
+    def run(self):
+        """Main viewer loop."""
+        # Get initial remote
+        self.current_remote = self.config.get_last_used_remote()
+
+        if not self.current_remote:
+            self.console.print("[red]No remotes configured.[/red]")
+            return
+
+        # Connection loop - allows retry after configuring remotes
+        while True:
+            # For Local remote, try to get API key from .env if not set
+            if self.current_remote["name"] == "Local" and not self.current_remote.get(
+                "api_key"
+            ):
+                env_key = self.config.get_api_key_from_env()
+                if env_key:
+                    self.current_remote["api_key"] = env_key
+
+            # Try to connect
+            with self.console.status("[bold]Connecting to proxy...", spinner="dots"):
+                stats = self.fetch_stats()
+
+            if stats is not None:
+                break  # Connected successfully
+
+            # Connection failed - show error with options
+            choice = self.show_connection_error()
+
+            if choice == "b":
+                return  # Exit to main menu
+            elif choice == "s":
+                self.show_switch_remote_screen()
+            elif choice == "m":
+                self.show_manage_remotes_screen()
+            elif choice == "r":
+                continue  # Retry connection
+
+            # After switch/manage, refresh current_remote from config
+            # (it may have been changed)
+            if self.current_remote:
+                updated = self.config.get_remote_by_name(self.current_remote["name"])
+                if updated:
+                    self.current_remote = updated
+
+        # Main loop
+        while self.running:
+            self.show_summary_screen()
+
+
+def run_quota_viewer():
+    """Entry point for the quota viewer."""
+    viewer = QuotaViewer()
+    viewer.run()
+
+
+if __name__ == "__main__":
+    run_quota_viewer()

src/proxy_app/quota_viewer_config.py

@@ -0,0 +1,300 @@
+# SPDX-License-Identifier: MIT
+# Copyright (c) 2026 Mirrowel
+
+"""
+Configuration management for the Quota Viewer.
+
+Handles remote proxy configurations including:
+- Multiple remote proxies (local, VPS, etc.)
+- API key storage per remote
+- Default and last-used remote tracking
+"""
+
+import json
+from pathlib import Path
+from typing import Any, Dict, List, Optional, Union
+
+
+class QuotaViewerConfig:
+    """Manages quota viewer configuration including remote proxies."""
+
+    def __init__(self, config_path: Optional[Path] = None):
+        """
+        Initialize the config manager.
+
+        Args:
+            config_path: Path to config file. Defaults to quota_viewer_config.json
+                        in the current directory or EXE directory.
+        """
+        if config_path is None:
+            import sys
+
+            if getattr(sys, "frozen", False):
+                base_dir = Path(sys.executable).parent
+            else:
+                base_dir = Path.cwd()
+            config_path = base_dir / "quota_viewer_config.json"
+
+        self.config_path = config_path
+        self.config = self._load()
+
+    def _load(self) -> Dict[str, Any]:
+        """Load config from file or return defaults."""
+        if self.config_path.exists():
+            try:
+                with open(self.config_path, "r", encoding="utf-8") as f:
+                    config = json.load(f)
+                # Ensure required fields exist
+                if "remotes" not in config:
+                    config["remotes"] = []
+                return config
+            except (json.JSONDecodeError, IOError):
+                pass
+
+        # Return default config with Local remote
+        return {
+            "remotes": [
+                {
+                    "name": "Local",
+                    "host": "127.0.0.1",
+                    "port": 8000,
+                    "api_key": None,
+                    "is_default": True,
+                }
+            ],
+            "last_used": "Local",
+        }
+
+    def _save(self) -> bool:
+        """Save config to file. Returns True on success."""
+        try:
+            with open(self.config_path, "w", encoding="utf-8") as f:
+                json.dump(self.config, f, indent=2)
+            return True
+        except IOError:
+            return False
+
+    def get_remotes(self) -> List[Dict[str, Any]]:
+        """Get list of all configured remotes."""
+        return self.config.get("remotes", [])
+
+    def get_remote_by_name(self, name: str) -> Optional[Dict[str, Any]]:
+        """Get a remote by name."""
+        for remote in self.config.get("remotes", []):
+            if remote["name"] == name:
+                return remote
+        return None
+
+    def get_default_remote(self) -> Optional[Dict[str, Any]]:
+        """Get the default remote."""
+        for remote in self.config.get("remotes", []):
+            if remote.get("is_default"):
+                return remote
+        # Fallback to first remote
+        remotes = self.config.get("remotes", [])
+        return remotes[0] if remotes else None
+
+    def get_last_used_remote(self) -> Optional[Dict[str, Any]]:
+        """Get the last used remote, or default if not set."""
+        last_used_name = self.config.get("last_used")
+        if last_used_name:
+            remote = self.get_remote_by_name(last_used_name)
+            if remote:
+                return remote
+        return self.get_default_remote()
+
+    def set_last_used(self, name: str) -> bool:
+        """Set the last used remote name."""
+        self.config["last_used"] = name
+        return self._save()
+
+    def add_remote(
+        self,
+        name: str,
+        host: str,
+        port: Optional[Union[int, str]] = 8000,
+        api_key: Optional[str] = None,
+        is_default: bool = False,
+    ) -> bool:
+        """
+        Add a new remote configuration.
+
+        Args:
+            name: Display name for the remote
+            host: Hostname, IP address, or full URL (e.g., https://api.example.com/v1)
+            port: Port number (default 8000). Can be None or empty string for full URLs.
+            api_key: Optional API key for authentication
+            is_default: Whether this should be the default remote
+
+        Returns:
+            True on success, False if name already exists
+        """
+        # Check for duplicate name
+        if self.get_remote_by_name(name):
+            return False
+
+        # If setting as default, clear default from others
+        if is_default:
+            for remote in self.config.get("remotes", []):
+                remote["is_default"] = False
+
+        # Normalize port - allow empty/None for full URL hosts
+        if port == "" or port is None:
+            normalized_port = ""
+        else:
+            normalized_port = (
+                int(port) if isinstance(port, str) and port.isdigit() else port
+            )
+
+        remote = {
+            "name": name,
+            "host": host,
+            "port": normalized_port,
+            "api_key": api_key,
+            "is_default": is_default,
+        }
+        self.config.setdefault("remotes", []).append(remote)
+        return self._save()
+
+    def update_remote(self, name: str, **kwargs) -> bool:
+        """
+        Update an existing remote configuration.
+
+        Args:
+            name: Name of the remote to update
+            **kwargs: Fields to update (host, port, api_key, is_default, new_name)
+                      port can be int, str, or empty string for full URL hosts
+
+        Returns:
+            True on success, False if remote not found
+        """
+        remote = self.get_remote_by_name(name)
+        if not remote:
+            return False
+
+        # Handle rename
+        if "new_name" in kwargs:
+            new_name = kwargs.pop("new_name")
+            if new_name != name and self.get_remote_by_name(new_name):
+                return False  # New name already exists
+            remote["name"] = new_name
+            # Update last_used if it was this remote
+            if self.config.get("last_used") == name:
+                self.config["last_used"] = new_name
+
+        # If setting as default, clear default from others
+        if kwargs.get("is_default"):
+            for r in self.config.get("remotes", []):
+                r["is_default"] = False
+
+        # Update other fields
+        for key in ("host", "port", "api_key", "is_default"):
+            if key in kwargs:
+                remote[key] = kwargs[key]
+
+        return self._save()
+
+    def delete_remote(self, name: str) -> bool:
+        """
+        Delete a remote configuration.
+
+        Args:
+            name: Name of the remote to delete
+
+        Returns:
+            True on success, False if remote not found or is the only one
+        """
+        remotes = self.config.get("remotes", [])
+        if len(remotes) <= 1:
+            return False  # Don't delete the last remote
+
+        for i, remote in enumerate(remotes):
+            if remote["name"] == name:
+                remotes.pop(i)
+                # Update last_used if it was this remote
+                if self.config.get("last_used") == name:
+                    self.config["last_used"] = remotes[0]["name"] if remotes else None
+                return self._save()
+        return False
+
+    def set_default_remote(self, name: str) -> bool:
+        """Set a remote as the default."""
+        remote = self.get_remote_by_name(name)
+        if not remote:
+            return False
+
+        # Clear default from all remotes
+        for r in self.config.get("remotes", []):
+            r["is_default"] = False
+
+        # Set new default
+        remote["is_default"] = True
+        return self._save()
+
+    def sync_with_launcher_config(self) -> None:
+        """
+        Sync the Local remote with launcher_config.json if it exists.
+
+        This ensures the Local remote always matches the launcher settings.
+        """
+        import sys
+
+        if getattr(sys, "frozen", False):
+            base_dir = Path(sys.executable).parent
+        else:
+            base_dir = Path.cwd()
+
+        launcher_config_path = base_dir / "launcher_config.json"
+
+        if launcher_config_path.exists():
+            try:
+                with open(launcher_config_path, "r", encoding="utf-8") as f:
+                    launcher_config = json.load(f)
+
+                host = launcher_config.get("host", "127.0.0.1")
+                port = launcher_config.get("port", 8000)
+
+                # Update Local remote
+                local_remote = self.get_remote_by_name("Local")
+                if local_remote:
+                    local_remote["host"] = host
+                    local_remote["port"] = port
+                    self._save()
+                else:
+                    # Create Local remote if it doesn't exist
+                    self.add_remote("Local", host, port, is_default=True)
+
+            except (json.JSONDecodeError, IOError):
+                pass
+
+    def get_api_key_from_env(self) -> Optional[str]:
+        """
+        Get PROXY_API_KEY from .env file for Local remote.
+
+        Returns:
+            API key string or None
+        """
+        import sys
+
+        if getattr(sys, "frozen", False):
+            base_dir = Path(sys.executable).parent
+        else:
+            base_dir = Path.cwd()
+
+        env_path = base_dir / ".env"
+        if not env_path.exists():
+            return None
+
+        try:
+            with open(env_path, "r", encoding="utf-8") as f:
+                for line in f:
+                    line = line.strip()
+                    if line.startswith("PROXY_API_KEY="):
+                        value = line.split("=", 1)[1].strip()
+                        # Remove quotes if present
+                        if value and value[0] in ('"', "'") and value[-1] == value[0]:
+                            value = value[1:-1]
+                        return value if value else None
+        except IOError:
+            pass
+        return None

src/proxy_app/request_logger.py

@@ -0,0 +1,34 @@
+# SPDX-License-Identifier: MIT
+# Copyright (c) 2026 Mirrowel
+
+import json
+import os
+from datetime import datetime
+from pathlib import Path
+import uuid
+from typing import Literal, Dict
+import logging
+
+from .provider_urls import get_provider_endpoint
+
+def log_request_to_console(url: str, headers: dict, client_info: tuple, request_data: dict):
+    """
+    Logs a concise, single-line summary of an incoming request to the console.
+    """
+    time_str = datetime.now().strftime("%H:%M")
+    model_full = request_data.get("model", "N/A")
+    
+    provider = "N/A"
+    model_name = model_full
+    endpoint_url = "N/A"
+
+    if '/' in model_full:
+        parts = model_full.split('/', 1)
+        provider = parts[0]
+        model_name = parts[1]
+        # Use the helper function to get the full endpoint URL
+        endpoint_url = get_provider_endpoint(provider, model_name, url) or "N/A"
+
+    log_message = f"{time_str} - {client_info[0]}:{client_info[1]} - provider: {provider}, model: {model_name} - {endpoint_url}"
+    logging.info(log_message)
+

src/rotator_library/COPYING

@@ -0,0 +1,674 @@
+                    GNU GENERAL PUBLIC LICENSE
+                       Version 3, 29 June 2007
+
+ Copyright (C) 2007 Free Software Foundation, Inc. <https://fsf.org/>
+ Everyone is permitted to copy and distribute verbatim copies
+ of this license document, but changing it is not allowed.
+
+                            Preamble
+
+  The GNU General Public License is a free, copyleft license for
+software and other kinds of works.
+
+  The licenses for most software and other practical works are designed
+to take away your freedom to share and change the works.  By contrast,
+the GNU General Public License is intended to guarantee your freedom to
+share and change all versions of a program--to make sure it remains free
+software for all its users.  We, the Free Software Foundation, use the
+GNU General Public License for most of our software; it applies also to
+any other work released this way by its authors.  You can apply it to
+your programs, too.
+
+  When we speak of free software, we are referring to freedom, not
+price.  Our General Public Licenses are designed to make sure that you
+have the freedom to distribute copies of free software (and charge for
+them if you wish), that you receive source code or can get it if you
+want it, that you can change the software or use pieces of it in new
+free programs, and that you know you can do these things.
+
+  To protect your rights, we need to prevent others from denying you
+these rights or asking you to surrender the rights.  Therefore, you have
+certain responsibilities if you distribute copies of the software, or if
+you modify it: responsibilities to respect the freedom of others.
+
+  For example, if you distribute copies of such a program, whether
+gratis or for a fee, you must pass on to the recipients the same
+freedoms that you received.  You must make sure that they, too, receive
+or can get the source code.  And you must show them these terms so they
+know their rights.
+
+  Developers that use the GNU GPL protect your rights with two steps:
+(1) assert copyright on the software, and (2) offer you this License
+giving you legal permission to copy, distribute and/or modify it.
+
+  For the developers' and authors' protection, the GPL clearly explains
+that there is no warranty for this free software.  For both users' and
+authors' sake, the GPL requires that modified versions be marked as
+changed, so that their problems will not be attributed erroneously to
+authors of previous versions.
+
+  Some devices are designed to deny users access to install or run
+modified versions of the software inside them, although the manufacturer
+can do so.  This is fundamentally incompatible with the aim of
+protecting users' freedom to change the software.  The systematic
+pattern of such abuse occurs in the area of products for individuals to
+use, which is precisely where it is most unacceptable.  Therefore, we
+have designed this version of the GPL to prohibit the practice for those
+products.  If such problems arise substantially in other domains, we
+stand ready to extend this provision to those domains in future versions
+of the GPL, as needed to protect the freedom of users.
+
+  Finally, every program is threatened constantly by software patents.
+States should not allow patents to restrict development and use of
+software on general-purpose computers, but in those that do, we wish to
+avoid the special danger that patents applied to a free program could
+make it effectively proprietary.  To prevent this, the GPL assures that
+patents cannot be used to render the program non-free.
+
+  The precise terms and conditions for copying, distribution and
+modification follow.
+
+                       TERMS AND CONDITIONS
+
+  0. Definitions.
+
+  "This License" refers to version 3 of the GNU General Public License.
+
+  "Copyright" also means copyright-like laws that apply to other kinds of
+works, such as semiconductor masks.
+
+  "The Program" refers to any copyrightable work licensed under this
+License.  Each licensee is addressed as "you".  "Licensees" and
+"recipients" may be individuals or organizations.
+
+  To "modify" a work means to copy from or adapt all or part of the work
+in a fashion requiring copyright permission, other than the making of an
+exact copy.  The resulting work is called a "modified version" of the
+earlier work or a work "based on" the earlier work.
+
+  A "covered work" means either the unmodified Program or a work based
+on the Program.
+
+  To "propagate" a work means to do anything with it that, without
+permission, would make you directly or secondarily liable for
+infringement under applicable copyright law, except executing it on a
+computer or modifying a private copy.  Propagation includes copying,
+distribution (with or without modification), making available to the
+public, and in some countries other activities as well.
+
+  To "convey" a work means any kind of propagation that enables other
+parties to make or receive copies.  Mere interaction with a user through
+a computer network, with no transfer of a copy, is not conveying.
+
+  An interactive user interface displays "Appropriate Legal Notices"
+to the extent that it includes a convenient and prominently visible
+feature that (1) displays an appropriate copyright notice, and (2)
+tells the user that there is no warranty for the work (except to the
+extent that warranties are provided), that licensees may convey the
+work under this License, and how to view a copy of this License.  If
+the interface presents a list of user commands or options, such as a
+menu, a prominent item in the list meets this criterion.
+
+  1. Source Code.
+
+  The "source code" for a work means the preferred form of the work
+for making modifications to it.  "Object code" means any non-source
+form of a work.
+
+  A "Standard Interface" means an interface that either is an official
+standard defined by a recognized standards body, or, in the case of
+interfaces specified for a particular programming language, one that
+is widely used among developers working in that language.
+
+  The "System Libraries" of an executable work include anything, other
+than the work as a whole, that (a) is included in the normal form of
+packaging a Major Component, but which is not part of that Major
+Component, and (b) serves only to enable use of the work with that
+Major Component, or to implement a Standard Interface for which an
+implementation is available to the public in source code form.  A
+"Major Component", in this context, means a major essential component
+(kernel, window system, and so on) of the specific operating system
+(if any) on which the executable work runs, or a compiler used to
+produce the work, or an object code interpreter used to run it.
+
+  The "Corresponding Source" for a work in object code form means all
+the source code needed to generate, install, and (for an executable
+work) run the object code and to modify the work, including scripts to
+control those activities.  However, it does not include the work's
+System Libraries, or general-purpose tools or generally available free
+programs which are used unmodified in performing those activities but
+which are not part of the work.  For example, Corresponding Source
+includes interface definition files associated with source files for
+the work, and the source code for shared libraries and dynamically
+linked subprograms that the work is specifically designed to require,
+such as by intimate data communication or control flow between those
+subprograms and other parts of the work.
+
+  The Corresponding Source need not include anything that users
+can regenerate automatically from other parts of the Corresponding
+Source.
+
+  The Corresponding Source for a work in source code form is that
+same work.
+
+  2. Basic Permissions.
+
+  All rights granted under this License are granted for the term of
+copyright on the Program, and are irrevocable provided the stated
+conditions are met.  This License explicitly affirms your unlimited
+permission to run the unmodified Program.  The output from running a
+covered work is covered by this License only if the output, given its
+content, constitutes a covered work.  This License acknowledges your
+rights of fair use or other equivalent, as provided by copyright law.
+
+  You may make, run and propagate covered works that you do not
+convey, without conditions so long as your license otherwise remains
+in force.  You may convey covered works to others for the sole purpose
+of having them make modifications exclusively for you, or provide you
+with facilities for running those works, provided that you comply with
+the terms of this License in conveying all material for which you do
+not control copyright.  Those thus making or running the covered works
+for you must do so exclusively on your behalf, under your direction
+and control, on terms that prohibit them from making any copies of
+your copyrighted material outside their relationship with you.
+
+  Conveying under any other circumstances is permitted solely under
+the conditions stated below.  Sublicensing is not allowed; section 10
+makes it unnecessary.
+
+  3. Protecting Users' Legal Rights From Anti-Circumvention Law.
+
+  No covered work shall be deemed part of an effective technological
+measure under any applicable law fulfilling obligations under article
+11 of the WIPO copyright treaty adopted on 20 December 1996, or
+similar laws prohibiting or restricting circumvention of such
+measures.
+
+  When you convey a covered work, you waive any legal power to forbid
+circumvention of technological measures to the extent such circumvention
+is effected by exercising rights under this License with respect to
+the covered work, and you disclaim any intention to limit operation or
+modification of the work as a means of enforcing, against the work's
+users, your or third parties' legal rights to forbid circumvention of
+technological measures.
+
+  4. Conveying Verbatim Copies.
+
+  You may convey verbatim copies of the Program's source code as you
+receive it, in any medium, provided that you conspicuously and
+appropriately publish on each copy an appropriate copyright notice;
+keep intact all notices stating that this License and any
+non-permissive terms added in accord with section 7 apply to the code;
+keep intact all notices of the absence of any warranty; and give all
+recipients a copy of this License along with the Program.
+
+  You may charge any price or no price for each copy that you convey,
+and you may offer support or warranty protection for a fee.
+
+  5. Conveying Modified Source Versions.
+
+  You may convey a work based on the Program, or the modifications to
+produce it from the Program, in the form of source code under the
+terms of section 4, provided that you also meet all of these conditions:
+
+    a) The work must carry prominent notices stating that you modified
+    it, and giving a relevant date.
+
+    b) The work must carry prominent notices stating that it is
+    released under this License and any conditions added under section
+    7.  This requirement modifies the requirement in section 4 to
+    "keep intact all notices".
+
+    c) You must license the entire work, as a whole, under this
+    License to anyone who comes into possession of a copy.  This
+    License will therefore apply, along with any applicable section 7
+    additional terms, to the whole of the work, and all its parts,
+    regardless of how they are packaged.  This License gives no
+    permission to license the work in any other way, but it does not
+    invalidate such permission if you have separately received it.
+
+    d) If the work has interactive user interfaces, each must display
+    Appropriate Legal Notices; however, if the Program has interactive
+    interfaces that do not display Appropriate Legal Notices, your
+    work need not make them do so.
+
+  A compilation of a covered work with other separate and independent
+works, which are not by their nature extensions of the covered work,
+and which are not combined with it such as to form a larger program,
+in or on a volume of a storage or distribution medium, is called an
+"aggregate" if the compilation and its resulting copyright are not
+used to limit the access or legal rights of the compilation's users
+beyond what the individual works permit.  Inclusion of a covered work
+in an aggregate does not cause this License to apply to the other
+parts of the aggregate.
+
+  6. Conveying Non-Source Forms.
+
+  You may convey a covered work in object code form under the terms
+of sections 4 and 5, provided that you also convey the
+machine-readable Corresponding Source under the terms of this License,
+in one of these ways:
+
+    a) Convey the object code in, or embodied in, a physical product
+    (including a physical distribution medium), accompanied by the
+    Corresponding Source fixed on a durable physical medium
+    customarily used for software interchange.
+
+    b) Convey the object code in, or embodied in, a physical product
+    (including a physical distribution medium), accompanied by a
+    written offer, valid for at least three years and valid for as
+    long as you offer spare parts or customer support for that product
+    model, to give anyone who possesses the object code either (1) a
+    copy of the Corresponding Source for all the software in the
+    product that is covered by this License, on a durable physical
+    medium customarily used for software interchange, for a price no
+    more than your reasonable cost of physically performing this
+    conveying of source, or (2) access to copy the
+    Corresponding Source from a network server at no charge.
+
+    c) Convey individual copies of the object code with a copy of the
+    written offer to provide the Corresponding Source.  This
+    alternative is allowed only occasionally and noncommercially, and
+    only if you received the object code with such an offer, in accord
+    with subsection 6b.
+
+    d) Convey the object code by offering access from a designated
+    place (gratis or for a charge), and offer equivalent access to the
+    Corresponding Source in the same way through the same place at no
+    further charge.  You need not require recipients to copy the
+    Corresponding Source along with the object code.  If the place to
+    copy the object code is a network server, the Corresponding Source
+    may be on a different server (operated by you or a third party)
+    that supports equivalent copying facilities, provided you maintain
+    clear directions next to the object code saying where to find the
+    Corresponding Source.  Regardless of what server hosts the
+    Corresponding Source, you remain obligated to ensure that it is
+    available for as long as needed to satisfy these requirements.
+
+    e) Convey the object code using peer-to-peer transmission, provided
+    you inform other peers where the object code and Corresponding
+    Source of the work are being offered to the general public at no
+    charge under subsection 6d.
+
+  A separable portion of the object code, whose source code is excluded
+from the Corresponding Source as a System Library, need not be
+included in conveying the object code work.
+
+  A "User Product" is either (1) a "consumer product", which means any
+tangible personal property which is normally used for personal, family,
+or household purposes, or (2) anything designed or sold for incorporation
+into a dwelling.  In determining whether a product is a consumer product,
+doubtful cases shall be resolved in favor of coverage.  For a particular
+product received by a particular user, "normally used" refers to a
+typical or common use of that class of product, regardless of the status
+of the particular user or of the way in which the particular user
+actually uses, or expects or is expected to use, the product.  A product
+is a consumer product regardless of whether the product has substantial
+commercial, industrial or non-consumer uses, unless such uses represent
+the only significant mode of use of the product.
+
+  "Installation Information" for a User Product means any methods,
+procedures, authorization keys, or other information required to install
+and execute modified versions of a covered work in that User Product from
+a modified version of its Corresponding Source.  The information must
+suffice to ensure that the continued functioning of the modified object
+code is in no case prevented or interfered with solely because
+modification has been made.
+
+  If you convey an object code work under this section in, or with, or
+specifically for use in, a User Product, and the conveying occurs as
+part of a transaction in which the right of possession and use of the
+User Product is transferred to the recipient in perpetuity or for a
+fixed term (regardless of how the transaction is characterized), the
+Corresponding Source conveyed under this section must be accompanied
+by the Installation Information.  But this requirement does not apply
+if neither you nor any third party retains the ability to install
+modified object code on the User Product (for example, the work has
+been installed in ROM).
+
+  The requirement to provide Installation Information does not include a
+requirement to continue to provide support service, warranty, or updates
+for a work that has been modified or installed by the recipient, or for
+the User Product in which it has been modified or installed.  Access to a
+network may be denied when the modification itself materially and
+adversely affects the operation of the network or violates the rules and
+protocols for communication across the network.
+
+  Corresponding Source conveyed, and Installation Information provided,
+in accord with this section must be in a format that is publicly
+documented (and with an implementation available to the public in
+source code form), and must require no special password or key for
+unpacking, reading or copying.
+
+  7. Additional Terms.
+
+  "Additional permissions" are terms that supplement the terms of this
+License by making exceptions from one or more of its conditions.
+Additional permissions that are applicable to the entire Program shall
+be treated as though they were included in this License, to the extent
+that they are valid under applicable law.  If additional permissions
+apply only to part of the Program, that part may be used separately
+under those permissions, but the entire Program remains governed by
+this License without regard to the additional permissions.
+
+  When you convey a copy of a covered work, you may at your option
+remove any additional permissions from that copy, or from any part of
+it.  (Additional permissions may be written to require their own
+removal in certain cases when you modify the work.)  You may place
+additional permissions on material, added by you to a covered work,
+for which you have or can give appropriate copyright permission.
+
+  Notwithstanding any other provision of this License, for material you
+add to a covered work, you may (if authorized by the copyright holders of
+that material) supplement the terms of this License with terms:
+
+    a) Disclaiming warranty or limiting liability differently from the
+    terms of sections 15 and 16 of this License; or
+
+    b) Requiring preservation of specified reasonable legal notices or
+    author attributions in that material or in the Appropriate Legal
+    Notices displayed by works containing it; or
+
+    c) Prohibiting misrepresentation of the origin of that material, or
+    requiring that modified versions of such material be marked in
+    reasonable ways as different from the original version; or
+
+    d) Limiting the use for publicity purposes of names of licensors or
+    authors of the material; or
+
+    e) Declining to grant rights under trademark law for use of some
+    trade names, trademarks, or service marks; or
+
+    f) Requiring indemnification of licensors and authors of that
+    material by anyone who conveys the material (or modified versions of
+    it) with contractual assumptions of liability to the recipient, for
+    any liability that these contractual assumptions directly impose on
+    those licensors and authors.
+
+  All other non-permissive additional terms are considered "further
+restrictions" within the meaning of section 10.  If the Program as you
+received it, or any part of it, contains a notice stating that it is
+governed by this License along with a term that is a further
+restriction, you may remove that term.  If a license document contains
+a further restriction but permits relicensing or conveying under this
+License, you may add to a covered work material governed by the terms
+of that license document, provided that the further restriction does
+not survive such relicensing or conveying.
+
+  If you add terms to a covered work in accord with this section, you
+must place, in the relevant source files, a statement of the
+additional terms that apply to those files, or a notice indicating
+where to find the applicable terms.
+
+  Additional terms, permissive or non-permissive, may be stated in the
+form of a separately written license, or stated as exceptions;
+the above requirements apply either way.
+
+  8. Termination.
+
+  You may not propagate or modify a covered work except as expressly
+provided under this License.  Any attempt otherwise to propagate or
+modify it is void, and will automatically terminate your rights under
+this License (including any patent licenses granted under the third
+paragraph of section 11).
+
+  However, if you cease all violation of this License, then your
+license from a particular copyright holder is reinstated (a)
+provisionally, unless and until the copyright holder explicitly and
+finally terminates your license, and (b) permanently, if the copyright
+holder fails to notify you of the violation by some reasonable means
+prior to 60 days after the cessation.
+
+  Moreover, your license from a particular copyright holder is
+reinstated permanently if the copyright holder notifies you of the
+violation by some reasonable means, this is the first time you have
+received notice of violation of this License (for any work) from that
+copyright holder, and you cure the violation prior to 30 days after
+your receipt of the notice.
+
+  Termination of your rights under this section does not terminate the
+licenses of parties who have received copies or rights from you under
+this License.  If your rights have been terminated and not permanently
+reinstated, you do not qualify to receive new licenses for the same
+material under section 10.
+
+  9. Acceptance Not Required for Having Copies.
+
+  You are not required to accept this License in order to receive or
+run a copy of the Program.  Ancillary propagation of a covered work
+occurring solely as a consequence of using peer-to-peer transmission
+to receive a copy likewise does not require acceptance.  However,
+nothing other than this License grants you permission to propagate or
+modify any covered work.  These actions infringe copyright if you do
+not accept this License.  Therefore, by modifying or propagating a
+covered work, you indicate your acceptance of this License to do so.
+
+  10. Automatic Licensing of Downstream Recipients.
+
+  Each time you convey a covered work, the recipient automatically
+receives a license from the original licensors, to run, modify and
+propagate that work, subject to this License.  You are not responsible
+for enforcing compliance by third parties with this License.
+
+  An "entity transaction" is a transaction transferring control of an
+organization, or substantially all assets of one, or subdividing an
+organization, or merging organizations.  If propagation of a covered
+work results from an entity transaction, each party to that
+transaction who receives a copy of the work also receives whatever
+licenses to the work the party's predecessor in interest had or could
+give under the previous paragraph, plus a right to possession of the
+Corresponding Source of the work from the predecessor in interest, if
+the predecessor has it or can get it with reasonable efforts.
+
+  You may not impose any further restrictions on the exercise of the
+rights granted or affirmed under this License.  For example, you may
+not impose a license fee, royalty, or other charge for exercise of
+rights granted under this License, and you may not initiate litigation
+(including a cross-claim or counterclaim in a lawsuit) alleging that
+any patent claim is infringed by making, using, selling, offering for
+sale, or importing the Program or any portion of it.
+
+  11. Patents.
+
+  A "contributor" is a copyright holder who authorizes use under this
+License of the Program or a work on which the Program is based.  The
+work thus licensed is called the contributor's "contributor version".
+
+  A contributor's "essential patent claims" are all patent claims
+owned or controlled by the contributor, whether already acquired or
+hereafter acquired, that would be infringed by some manner, permitted
+by this License, of making, using, or selling its contributor version,
+but do not include claims that would be infringed only as a
+consequence of further modification of the contributor version.  For
+purposes of this definition, "control" includes the right to grant
+patent sublicenses in a manner consistent with the requirements of
+this License.
+
+  Each contributor grants you a non-exclusive, worldwide, royalty-free
+patent license under the contributor's essential patent claims, to
+make, use, sell, offer for sale, import and otherwise run, modify and
+propagate the contents of its contributor version.
+
+  In the following three paragraphs, a "patent license" is any express
+agreement or commitment, however denominated, not to enforce a patent
+(such as an express permission to practice a patent or covenant not to
+sue for patent infringement).  To "grant" such a patent license to a
+party means to make such an agreement or commitment not to enforce a
+patent against the party.
+
+  If you convey a covered work, knowingly relying on a patent license,
+and the Corresponding Source of the work is not available for anyone
+to copy, free of charge and under the terms of this License, through a
+publicly available network server or other readily accessible means,
+then you must either (1) cause the Corresponding Source to be so
+available, or (2) arrange to deprive yourself of the benefit of the
+patent license for this particular work, or (3) arrange, in a manner
+consistent with the requirements of this License, to extend the patent
+license to downstream recipients.  "Knowingly relying" means you have
+actual knowledge that, but for the patent license, your conveying the
+covered work in a country, or your recipient's use of the covered work
+in a country, would infringe one or more identifiable patents in that
+country that you have reason to believe are valid.
+
+  If, pursuant to or in connection with a single transaction or
+arrangement, you convey, or propagate by procuring conveyance of, a
+covered work, and grant a patent license to some of the parties
+receiving the covered work authorizing them to use, propagate, modify
+or convey a specific copy of the covered work, then the patent license
+you grant is automatically extended to all recipients of the covered
+work and works based on it.
+
+  A patent license is "discriminatory" if it does not include within
+the scope of its coverage, prohibits the exercise of, or is
+conditioned on the non-exercise of one or more of the rights that are
+specifically granted under this License.  You may not convey a covered
+work if you are a party to an arrangement with a third party that is
+in the business of distributing software, under which you make payment
+to the third party based on the extent of your activity of conveying
+the work, and under which the third party grants, to any of the
+parties who would receive the covered work from you, a discriminatory
+patent license (a) in connection with copies of the covered work
+conveyed by you (or copies made from those copies), or (b) primarily
+for and in connection with specific products or compilations that
+contain the covered work, unless you entered into that arrangement,
+or that patent license was granted, prior to 28 March 2007.
+
+  Nothing in this License shall be construed as excluding or limiting
+any implied license or other defenses to infringement that may
+otherwise be available to you under applicable patent law.
+
+  12. No Surrender of Others' Freedom.
+
+  If conditions are imposed on you (whether by court order, agreement or
+otherwise) that contradict the conditions of this License, they do not
+excuse you from the conditions of this License.  If you cannot convey a
+covered work so as to satisfy simultaneously your obligations under this
+License and any other pertinent obligations, then as a consequence you may
+not convey it at all.  For example, if you agree to terms that obligate you
+to collect a royalty for further conveying from those to whom you convey
+the Program, the only way you could satisfy both those terms and this
+License would be to refrain entirely from conveying the Program.
+
+  13. Use with the GNU Affero General Public License.
+
+  Notwithstanding any other provision of this License, you have
+permission to link or combine any covered work with a work licensed
+under version 3 of the GNU Affero General Public License into a single
+combined work, and to convey the resulting work.  The terms of this
+License will continue to apply to the part which is the covered work,
+but the special requirements of the GNU Affero General Public License,
+section 13, concerning interaction through a network will apply to the
+combination as such.
+
+  14. Revised Versions of this License.
+
+  The Free Software Foundation may publish revised and/or new versions of
+the GNU General Public License from time to time.  Such new versions will
+be similar in spirit to the present version, but may differ in detail to
+address new problems or concerns.
+
+  Each version is given a distinguishing version number.  If the
+Program specifies that a certain numbered version of the GNU General
+Public License "or any later version" applies to it, you have the
+option of following the terms and conditions either of that numbered
+version or of any later version published by the Free Software
+Foundation.  If the Program does not specify a version number of the
+GNU General Public License, you may choose any version ever published
+by the Free Software Foundation.
+
+  If the Program specifies that a proxy can decide which future
+versions of the GNU General Public License can be used, that proxy's
+public statement of acceptance of a version permanently authorizes you
+to choose that version for the Program.
+
+  Later license versions may give you additional or different
+permissions.  However, no additional obligations are imposed on any
+author or copyright holder as a result of your choosing to follow a
+later version.
+
+  15. Disclaimer of Warranty.
+
+  THERE IS NO WARRANTY FOR THE PROGRAM, TO THE EXTENT PERMITTED BY
+APPLICABLE LAW.  EXCEPT WHEN OTHERWISE STATED IN WRITING THE COPYRIGHT
+HOLDERS AND/OR OTHER PARTIES PROVIDE THE PROGRAM "AS IS" WITHOUT WARRANTY
+OF ANY KIND, EITHER EXPRESSED OR IMPLIED, INCLUDING, BUT NOT LIMITED TO,
+THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
+PURPOSE.  THE ENTIRE RISK AS TO THE QUALITY AND PERFORMANCE OF THE PROGRAM
+IS WITH YOU.  SHOULD THE PROGRAM PROVE DEFECTIVE, YOU ASSUME THE COST OF
+ALL NECESSARY SERVICING, REPAIR OR CORRECTION.
+
+  16. Limitation of Liability.
+
+  IN NO EVENT UNLESS REQUIRED BY APPLICABLE LAW OR AGREED TO IN WRITING
+WILL ANY COPYRIGHT HOLDER, OR ANY OTHER PARTY WHO MODIFIES AND/OR CONVEYS
+THE PROGRAM AS PERMITTED ABOVE, BE LIABLE TO YOU FOR DAMAGES, INCLUDING ANY
+GENERAL, SPECIAL, INCIDENTAL OR CONSEQUENTIAL DAMAGES ARISING OUT OF THE
+USE OR INABILITY TO USE THE PROGRAM (INCLUDING BUT NOT LIMITED TO LOSS OF
+DATA OR DATA BEING RENDERED INACCURATE OR LOSSES SUSTAINED BY YOU OR THIRD
+PARTIES OR A FAILURE OF THE PROGRAM TO OPERATE WITH ANY OTHER PROGRAMS),
+EVEN IF SUCH HOLDER OR OTHER PARTY HAS BEEN ADVISED OF THE POSSIBILITY OF
+SUCH DAMAGES.
+
+  17. Interpretation of Sections 15 and 16.
+
+  If the disclaimer of warranty and limitation of liability provided
+above cannot be given local legal effect according to their terms,
+reviewing courts shall apply local law that most closely approximates
+an absolute waiver of all civil liability in connection with the
+Program, unless a warranty or assumption of liability accompanies a
+copy of the Program in return for a fee.
+
+                     END OF TERMS AND CONDITIONS
+
+            How to Apply These Terms to Your New Programs
+
+  If you develop a new program, and you want it to be of the greatest
+possible use to the public, the best way to achieve this is to make it
+free software which everyone can redistribute and change under these terms.
+
+  To do so, attach the following notices to the program.  It is safest
+to attach them to the start of each source file to most effectively
+state the exclusion of warranty; and each file should have at least
+the "copyright" line and a pointer to where the full notice is found.
+
+    <one line to give the program's name and a brief idea of what it does.>
+    Copyright (C) <year>  <name of author>
+
+    This program is free software: you can redistribute it and/or modify
+    it under the terms of the GNU General Public License as published by
+    the Free Software Foundation, either version 3 of the License, or
+    (at your option) any later version.
+
+    This program is distributed in the hope that it will be useful,
+    but WITHOUT ANY WARRANTY; without even the implied warranty of
+    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+    GNU General Public License for more details.
+
+    You should have received a copy of the GNU General Public License
+    along with this program.  If not, see <https://www.gnu.org/licenses/>.
+
+Also add information on how to contact you by electronic and paper mail.
+
+  If the program does terminal interaction, make it output a short
+notice like this when it starts in an interactive mode:
+
+    <program>  Copyright (C) <year>  <name of author>
+    This program comes with ABSOLUTELY NO WARRANTY; for details type `show w'.
+    This is free software, and you are welcome to redistribute it
+    under certain conditions; type `show c' for details.
+
+The hypothetical commands `show w' and `show c' should show the appropriate
+parts of the General Public License.  Of course, your program's commands
+might be different; for a GUI interface, you would use an "about box".
+
+  You should also get your employer (if you work as a programmer) or school,
+if any, to sign a "copyright disclaimer" for the program, if necessary.
+For more information on this, and how to apply and follow the GNU GPL, see
+<https://www.gnu.org/licenses/>.
+
+  The GNU General Public License does not permit incorporating your program
+into proprietary programs.  If your program is a subroutine library, you
+may consider it more useful to permit linking proprietary applications with
+the library.  If this is what you want to do, use the GNU Lesser General
+Public License instead of this License.  But first, please read
+<https://www.gnu.org/licenses/why-not-lgpl.html>.

src/rotator_library/COPYING.LESSER

@@ -0,0 +1,165 @@
+                   GNU LESSER GENERAL PUBLIC LICENSE
+                       Version 3, 29 June 2007
+
+ Copyright (C) 2007 Free Software Foundation, Inc. <https://fsf.org/>
+ Everyone is permitted to copy and distribute verbatim copies
+ of this license document, but changing it is not allowed.
+
+
+  This version of the GNU Lesser General Public License incorporates
+the terms and conditions of version 3 of the GNU General Public
+License, supplemented by the additional permissions listed below.
+
+  0. Additional Definitions.
+
+  As used herein, "this License" refers to version 3 of the GNU Lesser
+General Public License, and the "GNU GPL" refers to version 3 of the GNU
+General Public License.
+
+  "The Library" refers to a covered work governed by this License,
+other than an Application or a Combined Work as defined below.
+
+  An "Application" is any work that makes use of an interface provided
+by the Library, but which is not otherwise based on the Library.
+Defining a subclass of a class defined by the Library is deemed a mode
+of using an interface provided by the Library.
+
+  A "Combined Work" is a work produced by combining or linking an
+Application with the Library.  The particular version of the Library
+with which the Combined Work was made is also called the "Linked
+Version".
+
+  The "Minimal Corresponding Source" for a Combined Work means the
+Corresponding Source for the Combined Work, excluding any source code
+for portions of the Combined Work that, considered in isolation, are
+based on the Application, and not on the Linked Version.
+
+  The "Corresponding Application Code" for a Combined Work means the
+object code and/or source code for the Application, including any data
+and utility programs needed for reproducing the Combined Work from the
+Application, but excluding the System Libraries of the Combined Work.
+
+  1. Exception to Section 3 of the GNU GPL.
+
+  You may convey a covered work under sections 3 and 4 of this License
+without being bound by section 3 of the GNU GPL.
+
+  2. Conveying Modified Versions.
+
+  If you modify a copy of the Library, and, in your modifications, a
+facility refers to a function or data to be supplied by an Application
+that uses the facility (other than as an argument passed when the
+facility is invoked), then you may convey a copy of the modified
+version:
+
+   a) under this License, provided that you make a good faith effort to
+   ensure that, in the event an Application does not supply the
+   function or data, the facility still operates, and performs
+   whatever part of its purpose remains meaningful, or
+
+   b) under the GNU GPL, with none of the additional permissions of
+   this License applicable to that copy.
+
+  3. Object Code Incorporating Material from Library Header Files.
+
+  The object code form of an Application may incorporate material from
+a header file that is part of the Library.  You may convey such object
+code under terms of your choice, provided that, if the incorporated
+material is not limited to numerical parameters, data structure
+layouts and accessors, or small macros, inline functions and templates
+(ten or fewer lines in length), you do both of the following:
+
+   a) Give prominent notice with each copy of the object code that the
+   Library is used in it and that the Library and its use are
+   covered by this License.
+
+   b) Accompany the object code with a copy of the GNU GPL and this license
+   document.
+
+  4. Combined Works.
+
+  You may convey a Combined Work under terms of your choice that,
+taken together, effectively do not restrict modification of the
+portions of the Library contained in the Combined Work and reverse
+engineering for debugging such modifications, if you also do each of
+the following:
+
+   a) Give prominent notice with each copy of the Combined Work that
+   the Library is used in it and that the Library and its use are
+   covered by this License.
+
+   b) Accompany the Combined Work with a copy of the GNU GPL and this license
+   document.
+
+   c) For a Combined Work that displays copyright notices during
+   execution, include the copyright notice for the Library among
+   these notices, as well as a reference directing the user to the
+   copies of the GNU GPL and this license document.
+
+   d) Do one of the following:
+
+       0) Convey the Minimal Corresponding Source under the terms of this
+       License, and the Corresponding Application Code in a form
+       suitable for, and under terms that permit, the user to
+       recombine or relink the Application with a modified version of
+       the Linked Version to produce a modified Combined Work, in the
+       manner specified by section 6 of the GNU GPL for conveying
+       Corresponding Source.
+
+       1) Use a suitable shared library mechanism for linking with the
+       Library.  A suitable mechanism is one that (a) uses at run time
+       a copy of the Library already present on the user's computer
+       system, and (b) will operate properly with a modified version
+       of the Library that is interface-compatible with the Linked
+       Version.
+
+   e) Provide Installation Information, but only if you would otherwise
+   be required to provide such information under section 6 of the
+   GNU GPL, and only to the extent that such information is
+   necessary to install and execute a modified version of the
+   Combined Work produced by recombining or relinking the
+   Application with a modified version of the Linked Version. (If
+   you use option 4d0, the Installation Information must accompany
+   the Minimal Corresponding Source and Corresponding Application
+   Code. If you use option 4d1, you must provide the Installation
+   Information in the manner specified by section 6 of the GNU GPL
+   for conveying Corresponding Source.)
+
+  5. Combined Libraries.
+
+  You may place library facilities that are a work based on the
+Library side by side in a single library together with other library
+facilities that are not Applications and are not covered by this
+License, and convey such a combined library under terms of your
+choice, if you do both of the following:
+
+   a) Accompany the combined library with a copy of the same work based
+   on the Library, uncombined with any other library facilities,
+   conveyed under the terms of this License.
+
+   b) Give prominent notice with the combined library that part of it
+   is a work based on the Library, and explaining where to find the
+   accompanying uncombined form of the same work.
+
+  6. Revised Versions of the GNU Lesser General Public License.
+
+  The Free Software Foundation may publish revised and/or new versions
+of the GNU Lesser General Public License from time to time. Such new
+versions will be similar in spirit to the present version, but may
+differ in detail to address new problems or concerns.
+
+  Each version is given a distinguishing version number. If the
+Library as you received it specifies that a certain numbered version
+of the GNU Lesser General Public License "or any later version"
+applies to it, you have the option of following the terms and
+conditions either of that published version or of any later version
+published by the Free Software Foundation. If the Library as you
+received it does not specify a version number of the GNU Lesser
+General Public License, you may choose any version of the GNU Lesser
+General Public License ever published by the Free Software Foundation.
+
+  If the Library as you received it specifies that a proxy can decide
+whether future versions of the GNU Lesser General Public License shall
+apply, that proxy's public statement of acceptance of any version is
+permanent authorization for you to choose that version for the
+Library.

src/rotator_library/README.md

@@ -0,0 +1,345 @@
+# Resilience & API Key Management Library
+
+A robust, asynchronous, and thread-safe Python library for managing a pool of API keys. It is designed to be integrated into applications (such as the Universal LLM API Proxy included in this project) to provide a powerful layer of resilience and high availability when interacting with multiple LLM providers.
+
+## Key Features
+
+-   **Asynchronous by Design**: Built with `asyncio` and `httpx` for high-performance, non-blocking I/O.
+-   **Anthropic API Compatibility**: Built-in translation layer (`anthropic_compat`) enables Anthropic API clients (like Claude Code) to use any supported provider.
+-   **Advanced Concurrency Control**: A single API key can be used for multiple concurrent requests. By default, it supports concurrent requests to *different* models. With configuration (`MAX_CONCURRENT_REQUESTS_PER_KEY_<PROVIDER>`), it can also support multiple concurrent requests to the *same* model using the same key.
+-   **Smart Key Management**: Selects the optimal key for each request using a tiered, model-aware locking strategy to distribute load evenly and maximize availability.
+-   **Configurable Rotation Strategy**: Choose between deterministic least-used selection (perfect balance) or default weighted random selection (unpredictable, harder to fingerprint).
+-   **Deadline-Driven Requests**: A global timeout ensures that no request, including all retries and key selections, exceeds a specified time limit.
+-   **OAuth & API Key Support**: Built-in support for standard API keys and complex OAuth flows.
+    -   **Gemini CLI**: Full OAuth 2.0 web flow with automatic project discovery, free-tier onboarding, and credential prioritization (paid vs free tier).
+    -   **Antigravity**: Full OAuth 2.0 support for Gemini 3, Gemini 2.5, and Claude Sonnet 4.5 models with thought signature caching(Full support for Gemini 3 and Claude models). **First on the scene to provide full support for Gemini 3** via Antigravity with advanced features like thought signature caching and tool hallucination prevention.
+    -   **Qwen Code**: Device Code flow support.
+    -   **iFlow**: Authorization Code flow with local callback handling.
+-   **Stateless Deployment Ready**: Can load complex OAuth credentials from environment variables, eliminating the need for physical credential files in containerized environments.
+-   **Intelligent Error Handling**:
+    -   **Escalating Per-Model Cooldowns**: Failed keys are placed on a temporary, escalating cooldown for specific models.
+    -   **Key-Level Lockouts**: Keys failing across multiple models are temporarily removed from rotation.
+    -   **Stream Recovery**: The client detects mid-stream errors (like quota limits) and gracefully handles them.
+-   **Credential Prioritization**: Automatic tier detection and priority-based credential selection (e.g., paid tier credentials used first for models that require them).
+-   **Advanced Model Requirements**: Support for model-tier restrictions (e.g., Gemini 3 requires paid-tier credentials).
+-   **Robust Streaming Support**: Includes a wrapper for streaming responses that reassembles fragmented JSON chunks.
+-   **Detailed Usage Tracking**: Tracks daily and global usage for each key, persisted to a JSON file.
+-   **Automatic Daily Resets**: Automatically resets cooldowns and archives stats daily.
+-   **Provider Agnostic**: Works with any provider supported by `litellm`.
+-   **Extensible**: Easily add support for new providers through a simple plugin-based architecture.
+-   **Temperature Override**: Global temperature=0 override to prevent tool hallucination with low-temperature settings.
+-   **Shared OAuth Base**: Refactored OAuth implementation with reusable [`GoogleOAuthBase`](providers/google_oauth_base.py) for multiple providers.
+-   **Fair Cycle Rotation**: Ensures each credential exhausts at least once before any can be reused within a tier. Prevents a single credential from being repeatedly used while others sit idle. Configurable per provider with tracking modes and cross-tier support.
+-   **Custom Usage Caps**: Set custom limits per tier, per model/group that are more restrictive than actual API limits. Supports percentages (e.g., "80%") and multiple cooldown modes (`quota_reset`, `offset`, `fixed`). Credentials go on cooldown before hitting actual API limits.
+-   **Centralized Defaults**: All tunable defaults are defined in [`config/defaults.py`](config/defaults.py) for easy customization and visibility.
+
+## Installation
+
+To install the library, you can install it directly from a local path. Using the `-e` flag installs it in "editable" mode, which is recommended for development.
+
+```bash
+pip install -e .
+```
+
+## `RotatingClient` Class
+
+This is the main class for interacting with the library. It is designed to be a long-lived object that manages the state of your API key pool.
+
+### Initialization
+
+```python
+import os
+from dotenv import load_dotenv
+from rotator_library import RotatingClient
+
+# Load environment variables from .env file
+load_dotenv()
+
+# Dynamically load all provider API keys from environment variables
+api_keys = {}
+for key, value in os.environ.items():
+    # This pattern finds keys like "GEMINI_API_KEY_1" or "OPENAI_API_KEY"
+    if (key.endswith("_API_KEY") or "_API_KEY_" in key) and key != "PROXY_API_KEY":
+        # Extracts "gemini" from "GEMINI_API_KEY_1"
+        provider = key.split("_API_KEY")[0].lower()
+        if provider not in api_keys:
+            api_keys[provider] = []
+        api_keys[provider].append(value)
+
+# Initialize empty dictionary for OAuth credentials (or load from CredentialManager)
+oauth_credentials = {}
+
+client = RotatingClient(
+    api_keys=api_keys,
+    oauth_credentials=oauth_credentials,
+    max_retries=2,
+    usage_file_path="key_usage.json",
+    configure_logging=True,
+    global_timeout=30,
+    abort_on_callback_error=True,
+    litellm_provider_params={},
+    ignore_models={},
+    whitelist_models={},
+    enable_request_logging=False,
+    max_concurrent_requests_per_key={},
+    rotation_tolerance=2.0  # 0.0=deterministic, 2.0=recommended random
+)
+```
+
+#### Arguments
+
+-   `api_keys` (`Optional[Dict[str, List[str]]]`): A dictionary mapping provider names (e.g., "openai", "anthropic") to a list of API keys.
+-   `oauth_credentials` (`Optional[Dict[str, List[str]]]`): A dictionary mapping provider names (e.g., "gemini_cli", "qwen_code") to a list of file paths to OAuth credential JSON files.
+-   `max_retries` (`int`, default: `2`): The number of times to retry a request with the *same key* if a transient server error (e.g., 500, 503) occurs.
+-   `usage_file_path` (`str`, default: `"key_usage.json"`): The path to the JSON file where usage statistics (tokens, cost, success counts) are persisted.
+-   `configure_logging` (`bool`, default: `True`): If `True`, configures the library's logger to propagate logs to the root logger. Set to `False` if you want to handle logging configuration manually.
+-   `global_timeout` (`int`, default: `30`): A hard time limit (in seconds) for the entire request lifecycle. If the request (including all retries) takes longer than this, it is aborted.
+-   `abort_on_callback_error` (`bool`, default: `True`): If `True`, any exception raised by `pre_request_callback` will abort the request. If `False`, the error is logged and the request proceeds.
+-   `litellm_provider_params` (`Optional[Dict[str, Any]]`, default: `None`): A dictionary of extra parameters to pass to `litellm` for specific providers.
+-   `ignore_models` (`Optional[Dict[str, List[str]]]`, default: `None`): A dictionary where keys are provider names and values are lists of model names/patterns to exclude (blacklist). Supports wildcards (e.g., `"*-preview"`).
+-   `whitelist_models` (`Optional[Dict[str, List[str]]]`, default: `None`): A dictionary where keys are provider names and values are lists of model names/patterns to always include, overriding `ignore_models`.
+-   `enable_request_logging` (`bool`, default: `False`): If `True`, enables detailed per-request file logging (useful for debugging complex interactions).
+-   `max_concurrent_requests_per_key` (`Optional[Dict[str, int]]`, default: `None`): A dictionary defining the maximum number of concurrent requests allowed for a single API key for a specific provider. Defaults to 1 if not specified.
+-   `rotation_tolerance` (`float`, default: `0.0`): Controls credential rotation strategy:
+    - `0.0`: **Deterministic** - Always selects the least-used credential for perfect load balance.
+    - `2.0` (default, recommended): **Weighted Random** - Randomly selects credentials with bias toward less-used ones. Provides unpredictability (harder to fingerprint) while maintaining good balance.
+    - `5.0+`: **High Randomness** - Even heavily-used credentials have significant selection probability. Maximum unpredictability.
+    
+    The weight formula is: `weight = (max_usage - credential_usage) + tolerance + 1`
+    
+    **Use Cases:**
+    - `0.0`: When perfect load balance is critical
+    - `2.0`: When avoiding fingerprinting/rate limit detection is important
+    - `5.0+`: For stress testing or maximum unpredictability
+
+### Concurrency and Resource Management
+
+The `RotatingClient` is asynchronous and manages an `httpx.AsyncClient` internally. It's crucial to close the client properly to release resources. The recommended way is to use an `async with` block.
+
+```python
+import asyncio
+
+async def main():
+    async with RotatingClient(api_keys=api_keys) as client:
+        # ... use the client ...
+        response = await client.acompletion(
+            model="gemini/gemini-1.5-flash",
+            messages=[{"role": "user", "content": "Hello!"}]
+        )
+        print(response)
+
+asyncio.run(main())
+```
+
+### Methods
+
+#### `async def acompletion(self, **kwargs) -> Any:`
+
+This is the primary method for making API calls. It's a wrapper around `litellm.acompletion` that adds the core logic for key acquisition, selection, and retries.
+
+-   **Parameters**: Accepts the same keyword arguments as `litellm.acompletion`. The `model` parameter is required and must be a string in the format `provider/model_name`.
+-   **Returns**:
+    -   For non-streaming requests, it returns the `litellm` response object.
+    -   For streaming requests, it returns an async generator that yields OpenAI-compatible Server-Sent Events (SSE). The wrapper ensures that key locks are released and usage is recorded only after the stream is fully consumed.
+
+**Streaming Example:**
+
+```python
+async def stream_example():
+    async with RotatingClient(api_keys=api_keys) as client:
+        response_stream = await client.acompletion(
+            model="gemini/gemini-1.5-flash",
+            messages=[{"role": "user", "content": "Tell me a long story."}],
+            stream=True
+        )
+        async for chunk in response_stream:
+            print(chunk)
+
+asyncio.run(stream_example())
+```
+
+#### `async def aembedding(self, **kwargs) -> Any:`
+
+A wrapper around `litellm.aembedding` that provides the same key management and retry logic for embedding requests.
+
+#### `def token_count(self, model: str, text: str = None, messages: List[Dict[str, str]] = None) -> int:`
+
+Calculates the token count for a given text or list of messages using `litellm.token_counter`.
+
+#### `async def get_available_models(self, provider: str) -> List[str]:`
+
+Fetches a list of available models for a specific provider, applying any configured whitelists or blacklists. Results are cached in memory.
+
+#### `async def get_all_available_models(self, grouped: bool = True) -> Union[Dict[str, List[str]], List[str]]:`
+
+Fetches a dictionary of all available models, grouped by provider, or as a single flat list if `grouped=False`.
+
+#### `async def anthropic_messages(self, request, raw_request=None, pre_request_callback=None) -> Any:`
+
+Handle Anthropic Messages API requests. Accepts requests in Anthropic's format, translates them to OpenAI format internally, processes them through `acompletion`, and returns responses in Anthropic's format.
+
+-   **Parameters**:
+    -   `request`: An `AnthropicMessagesRequest` object (from `anthropic_compat.models`)
+    -   `raw_request`: Optional raw request object for client disconnect checks
+    -   `pre_request_callback`: Optional async callback before each API request
+-   **Returns**:
+    -   For non-streaming: dict in Anthropic Messages format
+    -   For streaming: AsyncGenerator yielding Anthropic SSE format strings
+
+#### `async def anthropic_count_tokens(self, request) -> dict:`
+
+Handle Anthropic count_tokens API requests. Counts the number of tokens that would be used by a Messages API request.
+
+-   **Parameters**: `request` - An `AnthropicCountTokensRequest` object
+-   **Returns**: Dict with `input_tokens` count in Anthropic format
+
+## Anthropic API Compatibility
+
+The library includes a translation layer (`anthropic_compat`) that enables Anthropic API clients to use any OpenAI-compatible provider.
+
+### Usage
+
+```python
+from rotator_library.anthropic_compat import (
+    AnthropicMessagesRequest,
+    AnthropicCountTokensRequest,
+    translate_anthropic_request,
+    openai_to_anthropic_response,
+    anthropic_streaming_wrapper,
+)
+
+# Create an Anthropic-format request
+request = AnthropicMessagesRequest(
+    model="gemini/gemini-2.5-flash",
+    max_tokens=1024,
+    messages=[{"role": "user", "content": "Hello!"}]
+)
+
+# Use with RotatingClient
+async with RotatingClient(api_keys=api_keys) as client:
+    response = await client.anthropic_messages(request)
+    print(response["content"][0]["text"])
+```
+
+### Features
+
+-   **Full Message Translation**: Converts between Anthropic and OpenAI message formats including text, images, tool_use, and tool_result blocks
+-   **Extended Thinking Support**: Translates Anthropic's `thinking` configuration to `reasoning_effort` for providers that support it
+-   **Streaming SSE Conversion**: Converts OpenAI streaming chunks to Anthropic's SSE event format (`message_start`, `content_block_delta`, etc.)
+-   **Cache Token Handling**: Properly translates `prompt_tokens_details.cached_tokens` to Anthropic's `cache_read_input_tokens`
+-   **Tool Call Support**: Full support for tool definitions and tool use/result blocks
+
+## Credential Tool
+
+The library includes a utility to manage credentials easily:
+
+```bash
+python -m src.rotator_library.credential_tool
+```
+
+Use this tool to:
+1.  **Initialize OAuth**: Run the interactive login flows for Gemini, Qwen, and iFlow.
+2.  **Export Credentials**: Generate `.env` compatible configuration blocks from your saved OAuth JSON files. This is essential for setting up stateless deployments.
+
+## Provider Specifics
+
+### Qwen Code
+-   **Auth**: Uses OAuth 2.0 Device Flow. Requires manual entry of email/identifier if not returned by the provider.
+-   **Resilience**: Injects a dummy tool (`do_not_call_me`) into requests with no tools to prevent known stream corruption issues on the API.
+-   **Reasoning**: Parses `<think>` tags in the response and exposes them as `reasoning_content`.
+-   **Schema Cleaning**: Recursively removes `strict` and `additionalProperties` from all tool schemas. Qwen's API has stricter validation than OpenAI's, and these properties cause `400 Bad Request` errors.
+
+### iFlow
+-   **Auth**: Uses Authorization Code Flow with a local callback server (port 11451).
+-   **Key Separation**: Distinguishes between the OAuth `access_token` (used to fetch user info) and the `api_key` (used for actual chat requests).
+-   **Resilience**: Similar to Qwen, injects a placeholder tool to stabilize streaming for empty tool lists.
+-   **Schema Cleaning**: Recursively removes `strict` and `additionalProperties` from all tool schemas to prevent API validation errors.
+-   **Custom Models**: Supports model definitions via `IFLOW_MODELS` environment variable (JSON array of model IDs or objects).
+
+### NVIDIA NIM
+-   **Discovery**: Dynamically fetches available models from the NVIDIA API.
+-   **Thinking**: Automatically injects the `thinking` parameter into `extra_body` for DeepSeek models (`deepseek-v3.1`, etc.) when `reasoning_effort` is set to low/medium/high.
+
+### Google Gemini (CLI)
+-   **Auth**: Simulates the Google Cloud CLI authentication flow.
+-   **Project Discovery**: Automatically discovers the default Google Cloud Project ID with enhanced onboarding flow.
+-   **Credential Prioritization**: Automatic detection and prioritization of paid vs free tier credentials.
+-   **Model Tier Requirements**: Gemini 3 models automatically filtered to paid-tier credentials only.
+-   **Gemini 3 Support**: Full support for Gemini 3 models with:
+    - `thinkingLevel` configuration (low/high)
+    - Tool hallucination prevention via system instruction injection
+    - ThoughtSignature caching for multi-turn conversations
+    - Parameter signature injection into tool descriptions
+-   **Rate Limits**: Implements smart fallback strategies (e.g., switching from `gemini-1.5-pro` to `gemini-1.5-pro-002`) when rate limits are hit.
+
+### Antigravity
+-   **Auth**: Uses OAuth 2.0 flow similar to Gemini CLI, with Antigravity-specific credentials and scopes.
+-   **Credential Prioritization**: Automatic detection and prioritization of paid vs free tier credentials (paid tier resets every 5 hours, free tier resets weekly).
+-   **Models**: Supports Gemini 3 Pro, Gemini 2.5 Flash/Flash Lite, Claude Sonnet 4.5 (with/without thinking), Claude Opus 4.5 (thinking only), and GPT-OSS 120B via Google's internal Antigravity API.
+-   **Quota Groups**: Models that share quota are automatically grouped:
+    - Claude/GPT-OSS: `claude-sonnet-4-5`, `claude-opus-4-5`, `gpt-oss-120b-medium`
+    - Gemini 3 Pro: `gemini-3-pro-high`, `gemini-3-pro-low`, `gemini-3-pro-preview`
+    - Gemini 2.5 Flash: `gemini-2.5-flash`, `gemini-2.5-flash-thinking`, `gemini-2.5-flash-lite`
+    - All models in a group deplete the usage of the group equally. So in claude group - it is beneficial to use only Opus, and forget about Sonnet and GPT-OSS.
+-   **Quota Baseline Tracking**: Background job fetches quota status from API every 5 minutes to provide accurate remaining quota estimates.
+-   **Thought Signature Caching**: Server-side caching of `thoughtSignature` data for multi-turn conversations with Gemini 3 models.
+-   **Tool Hallucination Prevention**: Automatic injection of system instructions and parameter signatures for Gemini 3 and Claude to prevent tool parameter hallucination.
+-   **Parallel Tool Usage Instruction**: Configurable instruction injection to encourage parallel tool calls (enabled by default for Claude).
+-   **Thinking Support**:
+    - Gemini 3: Uses `thinkingLevel` (string: "low"/"high")
+    - Gemini 2.5 Flash: Uses `-thinking` variant when `reasoning_effort` is provided
+    - Claude Sonnet 4.5: Uses `thinkingBudget` (optional - supports both thinking and non-thinking modes)
+    - Claude Opus 4.5: Uses `thinkingBudget` (always uses thinking variant)
+-   **Base URL Fallback**: Automatic fallback between sandbox and production endpoints.
+-   **Fair Cycle Rotation**: Enabled by default in sequential mode. Ensures all credentials cycle before reuse.
+-   **Custom Caps**: Configurable per-tier caps with offset cooldowns for pacing usage. See `config/defaults.py`.
+
+## Error Handling and Cooldowns
+
+The client uses a sophisticated error handling mechanism:
+
+-   **Error Classification**: All exceptions from `litellm` are passed through a `classify_error` function to determine their type (`rate_limit`, `authentication`, `server_error`, `quota`, `context_length`, etc.).
+-   **Server Errors**: The client will retry the request with the *same key* up to `max_retries` times, using an exponential backoff strategy.
+-   **Key-Specific Errors (Authentication, Quota, etc.)**: The client records the failure in the `UsageManager`, which applies an escalating cooldown to the key for that specific model. The client then immediately acquires a new key and continues its attempt to complete the request.
+-   **Escalating Cooldown Strategy**: Consecutive failures for a key on the same model result in increasing cooldown períods:
+    - 1st failure: 10 seconds
+    - 2nd failure: 30 seconds
+    - 3rd failure: 60 seconds
+    - 4th+ failure: 120 seconds
+-   **Key-Level Lockouts**: If a key fails on multiple different models (3+ distinct models), the `UsageManager` applies a global 5-minute lockout for that key, removing it from rotation entirely.
+-   **Authentication Errors**: Immediate 5-minute global lockout (key is assumed revoked or invalid).
+
+### Global Timeout and Deadline-Driven Logic
+
+To ensure predictable performance, the client now operates on a strict time budget defined by the `global_timeout` parameter.
+
+-   **Deadline Enforcement**: When a request starts, a `deadline` is set. The entire process, including all key rotations and retries, must complete before this deadline.
+-   **Deadline-Aware Retries**: If a retry requires a wait time that would exceed the remaining budget, the wait is skipped, and the client immediately rotates to the next key.
+-   **Silent Internal Errors**: Intermittent failures like provider capacity limits or temporary server errors are logged internally but are **not raised** to the caller. The client will simply rotate to the next key.
+
+## Extending with Provider Plugins
+
+The library uses a dynamic plugin system. To add support for a new provider's model list, you only need to:
+
+1.  **Create a new provider file** in `src/rotator_library/providers/` (e.g., `my_provider.py`).
+2.  **Implement the `ProviderInterface`**: Inside your new file, create a class that inherits from `ProviderInterface` and implements the `get_models` method.
+
+```python
+# src/rotator_library/providers/my_provider.py
+from .provider_interface import ProviderInterface
+from typing import List
+import httpx
+
+class MyProvider(ProviderInterface):
+    async def get_models(self, credential: str, client: httpx.AsyncClient) -> List[str]:
+        # Logic to fetch and return a list of model names
+        # The credential argument allows using the key to fetch models
+        pass
+```
+
+The system will automatically discover and register your new provider.
+
+## Detailed Documentation
+
+For a more in-depth technical explanation of the library's architecture, including the `UsageManager`'s concurrency model and the error classification system, please refer to the [Technical Documentation](../../DOCUMENTATION.md).
+

src/rotator_library/__init__.py

@@ -0,0 +1,48 @@
+# SPDX-License-Identifier: LGPL-3.0-only
+# Copyright (c) 2026 Mirrowel
+
+from typing import TYPE_CHECKING, Dict, Type
+
+from .client import RotatingClient
+
+# For type checkers (Pylint, mypy), import PROVIDER_PLUGINS statically
+# At runtime, it's lazy-loaded via __getattr__
+if TYPE_CHECKING:
+    from .providers import PROVIDER_PLUGINS
+    from .providers.provider_interface import ProviderInterface
+    from .model_info_service import ModelInfoService, ModelInfo, ModelMetadata
+    from . import anthropic_compat
+
+__all__ = [
+    "RotatingClient",
+    "PROVIDER_PLUGINS",
+    "ModelInfoService",
+    "ModelInfo",
+    "ModelMetadata",
+    "anthropic_compat",
+]
+
+
+def __getattr__(name):
+    """Lazy-load PROVIDER_PLUGINS, ModelInfoService, and anthropic_compat to speed up module import."""
+    if name == "PROVIDER_PLUGINS":
+        from .providers import PROVIDER_PLUGINS
+
+        return PROVIDER_PLUGINS
+    if name == "ModelInfoService":
+        from .model_info_service import ModelInfoService
+
+        return ModelInfoService
+    if name == "ModelInfo":
+        from .model_info_service import ModelInfo
+
+        return ModelInfo
+    if name == "ModelMetadata":
+        from .model_info_service import ModelMetadata
+
+        return ModelMetadata
+    if name == "anthropic_compat":
+        from . import anthropic_compat
+
+        return anthropic_compat
+    raise AttributeError(f"module {__name__!r} has no attribute {name!r}")

src/rotator_library/__pycache__/client.cpython-311.pyc

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:edc8f881bbdcc158873de4f9d619a679c2e38a0dfdfacafdb9b664340c8a4bad
+size 133767

src/rotator_library/__pycache__/client.cpython-314.pyc

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:78e2287189c4fa734c92541a15808197dba901d8a33dff8617180acec249d4f3
+size 130115

src/rotator_library/__pycache__/credential_tool.cpython-311.pyc

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6c2364484c477fa560b864a797018922635c7edf7949211a5fccf1e2204bf9a8
+size 115435

src/rotator_library/__pycache__/usage_manager.cpython-311.pyc

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cb69d1b0cb42c68574e65fb88bfb8c9df3da92145cc8a2dcb5158fded655b8e8
+size 156427

src/rotator_library/anthropic_compat/__init__.py

@@ -0,0 +1,70 @@
+# SPDX-License-Identifier: LGPL-3.0-only
+# Copyright (c) 2026 Mirrowel
+
+"""
+Anthropic API compatibility module for rotator_library.
+
+This module provides format translation between Anthropic's Messages API
+and OpenAI's Chat Completions API, enabling any OpenAI-compatible provider
+to work with Anthropic clients like Claude Code.
+
+Usage:
+    from rotator_library.anthropic_compat import (
+        AnthropicMessagesRequest,
+        AnthropicMessagesResponse,
+        translate_anthropic_request,
+        openai_to_anthropic_response,
+        anthropic_streaming_wrapper,
+    )
+"""
+
+from .models import (
+    AnthropicTextBlock,
+    AnthropicImageSource,
+    AnthropicImageBlock,
+    AnthropicToolUseBlock,
+    AnthropicToolResultBlock,
+    AnthropicMessage,
+    AnthropicTool,
+    AnthropicThinkingConfig,
+    AnthropicMessagesRequest,
+    AnthropicUsage,
+    AnthropicMessagesResponse,
+    AnthropicCountTokensRequest,
+    AnthropicCountTokensResponse,
+)
+
+from .translator import (
+    anthropic_to_openai_messages,
+    anthropic_to_openai_tools,
+    anthropic_to_openai_tool_choice,
+    openai_to_anthropic_response,
+    translate_anthropic_request,
+)
+
+from .streaming import anthropic_streaming_wrapper
+
+__all__ = [
+    # Models
+    "AnthropicTextBlock",
+    "AnthropicImageSource",
+    "AnthropicImageBlock",
+    "AnthropicToolUseBlock",
+    "AnthropicToolResultBlock",
+    "AnthropicMessage",
+    "AnthropicTool",
+    "AnthropicThinkingConfig",
+    "AnthropicMessagesRequest",
+    "AnthropicUsage",
+    "AnthropicMessagesResponse",
+    "AnthropicCountTokensRequest",
+    "AnthropicCountTokensResponse",
+    # Translator functions
+    "anthropic_to_openai_messages",
+    "anthropic_to_openai_tools",
+    "anthropic_to_openai_tool_choice",
+    "openai_to_anthropic_response",
+    "translate_anthropic_request",
+    # Streaming
+    "anthropic_streaming_wrapper",
+]

src/rotator_library/anthropic_compat/models.py

@@ -0,0 +1,147 @@
+# SPDX-License-Identifier: LGPL-3.0-only
+# Copyright (c) 2026 Mirrowel
+
+"""
+Pydantic models for the Anthropic Messages API.
+
+These models define the request and response formats for Anthropic's Messages API,
+enabling compatibility with Claude Code and other Anthropic API clients.
+"""
+
+from typing import Any, List, Optional, Union
+from pydantic import BaseModel
+
+
+# --- Content Blocks ---
+class AnthropicTextBlock(BaseModel):
+    """Anthropic text content block."""
+
+    type: str = "text"
+    text: str
+
+
+class AnthropicImageSource(BaseModel):
+    """Anthropic image source for base64 images."""
+
+    type: str = "base64"
+    media_type: str
+    data: str
+
+
+class AnthropicImageBlock(BaseModel):
+    """Anthropic image content block."""
+
+    type: str = "image"
+    source: AnthropicImageSource
+
+
+class AnthropicToolUseBlock(BaseModel):
+    """Anthropic tool use content block."""
+
+    type: str = "tool_use"
+    id: str
+    name: str
+    input: dict
+
+
+class AnthropicToolResultBlock(BaseModel):
+    """Anthropic tool result content block."""
+
+    type: str = "tool_result"
+    tool_use_id: str
+    content: Union[str, List[Any]]
+    is_error: Optional[bool] = None
+
+
+# --- Message and Tool Definitions ---
+class AnthropicMessage(BaseModel):
+    """Anthropic message format."""
+
+    role: str
+    content: Union[
+        str,
+        List[
+            Union[
+                AnthropicTextBlock,
+                AnthropicImageBlock,
+                AnthropicToolUseBlock,
+                AnthropicToolResultBlock,
+                dict,
+            ]
+        ],
+    ]
+
+
+class AnthropicTool(BaseModel):
+    """Anthropic tool definition."""
+
+    name: str
+    description: Optional[str] = None
+    input_schema: dict
+
+
+class AnthropicThinkingConfig(BaseModel):
+    """Anthropic thinking configuration."""
+
+    type: str  # "enabled" or "disabled"
+    budget_tokens: Optional[int] = None
+
+
+# --- Messages Request ---
+class AnthropicMessagesRequest(BaseModel):
+    """Anthropic Messages API request format."""
+
+    model: str
+    messages: List[AnthropicMessage]
+    max_tokens: int
+    system: Optional[Union[str, List[dict]]] = None
+    temperature: Optional[float] = None
+    top_p: Optional[float] = None
+    top_k: Optional[int] = None
+    stop_sequences: Optional[List[str]] = None
+    stream: Optional[bool] = False
+    tools: Optional[List[AnthropicTool]] = None
+    tool_choice: Optional[dict] = None
+    metadata: Optional[dict] = None
+    thinking: Optional[AnthropicThinkingConfig] = None
+
+
+# --- Messages Response ---
+class AnthropicUsage(BaseModel):
+    """Anthropic usage statistics."""
+
+    input_tokens: int
+    output_tokens: int
+    cache_creation_input_tokens: Optional[int] = None
+    cache_read_input_tokens: Optional[int] = None
+
+
+class AnthropicMessagesResponse(BaseModel):
+    """Anthropic Messages API response format."""
+
+    id: str
+    type: str = "message"
+    role: str = "assistant"
+    content: List[Union[AnthropicTextBlock, AnthropicToolUseBlock, dict]]
+    model: str
+    stop_reason: Optional[str] = None
+    stop_sequence: Optional[str] = None
+    usage: AnthropicUsage
+
+
+# --- Count Tokens ---
+class AnthropicCountTokensRequest(BaseModel):
+    """Anthropic count_tokens API request format."""
+
+    model: str
+    messages: List[AnthropicMessage]
+    system: Optional[Union[str, List[dict]]] = None
+    tools: Optional[List[AnthropicTool]] = None
+    tool_choice: Optional[dict] = None
+    thinking: Optional[AnthropicThinkingConfig] = None
+
+
+class AnthropicCountTokensResponse(BaseModel):
+    """Anthropic count_tokens API response format."""
+
+    input_tokens: int

src/rotator_library/anthropic_compat/streaming.py

@@ -0,0 +1,433 @@
+# SPDX-License-Identifier: LGPL-3.0-only
+# Copyright (c) 2026 Mirrowel
+
+"""
+Streaming wrapper for converting OpenAI streaming format to Anthropic streaming format.
+
+This module provides a framework-agnostic streaming wrapper that converts
+OpenAI SSE (Server-Sent Events) format to Anthropic's streaming format.
+"""
+
+import json
+import logging
+import uuid
+from typing import AsyncGenerator, Callable, Optional, Awaitable, Any, TYPE_CHECKING
+
+if TYPE_CHECKING:
+    from ..transaction_logger import TransactionLogger
+
+logger = logging.getLogger("rotator_library.anthropic_compat")
+
+
+async def anthropic_streaming_wrapper(
+    openai_stream: AsyncGenerator[str, None],
+    original_model: str,
+    request_id: Optional[str] = None,
+    is_disconnected: Optional[Callable[[], Awaitable[bool]]] = None,
+    transaction_logger: Optional["TransactionLogger"] = None,
+) -> AsyncGenerator[str, None]:
+    """
+    Convert OpenAI streaming format to Anthropic streaming format.
+
+    This is a framework-agnostic wrapper that can be used with any async web framework.
+    Instead of taking a FastAPI Request object, it accepts an optional callback function
+    to check for client disconnection.
+
+    Anthropic SSE events:
+    - message_start: Initial message metadata
+    - content_block_start: Start of a content block
+    - content_block_delta: Content chunk
+    - content_block_stop: End of a content block
+    - message_delta: Final message metadata (stop_reason, usage)
+    - message_stop: End of message
+
+    Args:
+        openai_stream: AsyncGenerator yielding OpenAI SSE format strings
+        original_model: The model name to include in responses
+        request_id: Optional request ID (auto-generated if not provided)
+        is_disconnected: Optional async callback that returns True if client disconnected
+        transaction_logger: Optional TransactionLogger for logging the final Anthropic response
+
+    Yields:
+        SSE format strings in Anthropic's streaming format
+    """
+    if request_id is None:
+        request_id = f"msg_{uuid.uuid4().hex[:24]}"
+
+    message_started = False
+    content_block_started = False
+    thinking_block_started = False
+    current_block_index = 0
+    tool_calls_by_index = {}  # Track tool calls by their index
+    tool_block_indices = {}  # Track which block index each tool call uses
+    input_tokens = 0
+    output_tokens = 0
+    cached_tokens = 0  # Track cached tokens for proper Anthropic format
+    accumulated_text = ""  # Track accumulated text for logging
+    accumulated_thinking = ""  # Track accumulated thinking for logging
+    stop_reason_final = "end_turn"  # Track final stop reason for logging
+
+    try:
+        async for chunk_str in openai_stream:
+            # Check for client disconnection if callback provided
+            if is_disconnected is not None and await is_disconnected():
+                break
+
+            if not chunk_str.strip() or not chunk_str.startswith("data:"):
+                continue
+
+            data_content = chunk_str[len("data:") :].strip()
+            if data_content == "[DONE]":
+                # CRITICAL: Send message_start if we haven't yet (e.g., empty response)
+                # Claude Code and other clients require message_start before message_stop
+                if not message_started:
+                    # Build usage with cached tokens properly handled
+                    usage_dict = {
+                        "input_tokens": input_tokens - cached_tokens,
+                        "output_tokens": 0,
+                    }
+                    if cached_tokens > 0:
+                        usage_dict["cache_read_input_tokens"] = cached_tokens
+                        usage_dict["cache_creation_input_tokens"] = 0
+
+                    message_start = {
+                        "type": "message_start",
+                        "message": {
+                            "id": request_id,
+                            "type": "message",
+                            "role": "assistant",
+                            "content": [],
+                            "model": original_model,
+                            "stop_reason": None,
+                            "stop_sequence": None,
+                            "usage": usage_dict,
+                        },
+                    }
+                    yield f"event: message_start\ndata: {json.dumps(message_start)}\n\n"
+                    message_started = True
+
+                # Close any open thinking block
+                if thinking_block_started:
+                    yield f'event: content_block_stop\ndata: {{"type": "content_block_stop", "index": {current_block_index}}}\n\n'
+                    current_block_index += 1
+                    thinking_block_started = False
+
+                # Close any open text block
+                if content_block_started:
+                    yield f'event: content_block_stop\ndata: {{"type": "content_block_stop", "index": {current_block_index}}}\n\n'
+                    current_block_index += 1
+                    content_block_started = False
+
+                # Close all open tool_use blocks
+                for tc_index in sorted(tool_block_indices.keys()):
+                    block_idx = tool_block_indices[tc_index]
+                    yield f'event: content_block_stop\ndata: {{"type": "content_block_stop", "index": {block_idx}}}\n\n'
+
+                # Determine stop_reason based on whether we had tool calls
+                stop_reason = "tool_use" if tool_calls_by_index else "end_turn"
+                stop_reason_final = stop_reason
+
+                # Build final usage dict with cached tokens
+                final_usage = {"output_tokens": output_tokens}
+                if cached_tokens > 0:
+                    final_usage["cache_read_input_tokens"] = cached_tokens
+                    final_usage["cache_creation_input_tokens"] = 0
+
+                # Send message_delta with final info
+                yield f'event: message_delta\ndata: {{"type": "message_delta", "delta": {{"stop_reason": "{stop_reason}", "stop_sequence": null}}, "usage": {json.dumps(final_usage)}}}\n\n'
+
+                # Send message_stop
+                yield 'event: message_stop\ndata: {"type": "message_stop"}\n\n'
+
+                # Log final Anthropic response if logger provided
+                if transaction_logger:
+                    # Build content blocks for logging
+                    content_blocks = []
+                    if accumulated_thinking:
+                        content_blocks.append(
+                            {
+                                "type": "thinking",
+                                "thinking": accumulated_thinking,
+                            }
+                        )
+                    if accumulated_text:
+                        content_blocks.append(
+                            {
+                                "type": "text",
+                                "text": accumulated_text,
+                            }
+                        )
+                    # Add tool use blocks
+                    for tc_index in sorted(tool_calls_by_index.keys()):
+                        tc = tool_calls_by_index[tc_index]
+                        # Parse arguments JSON string to dict
+                        try:
+                            input_data = json.loads(tc.get("arguments", "{}"))
+                        except json.JSONDecodeError:
+                            input_data = {}
+                        content_blocks.append(
+                            {
+                                "type": "tool_use",
+                                "id": tc.get("id", ""),
+                                "name": tc.get("name", ""),
+                                "input": input_data,
+                            }
+                        )
+
+                    # Build usage for logging
+                    log_usage = {
+                        "input_tokens": input_tokens - cached_tokens,
+                        "output_tokens": output_tokens,
+                    }
+                    if cached_tokens > 0:
+                        log_usage["cache_read_input_tokens"] = cached_tokens
+                        log_usage["cache_creation_input_tokens"] = 0
+
+                    anthropic_response = {
+                        "id": request_id,
+                        "type": "message",
+                        "role": "assistant",
+                        "content": content_blocks,
+                        "model": original_model,
+                        "stop_reason": stop_reason_final,
+                        "stop_sequence": None,
+                        "usage": log_usage,
+                    }
+                    transaction_logger.log_response(
+                        anthropic_response,
+                        filename="anthropic_response.json",
+                    )
+
+                break
+
+            try:
+                chunk = json.loads(data_content)
+            except json.JSONDecodeError:
+                continue
+
+            # Extract usage if present
+            # Note: Google's promptTokenCount INCLUDES cached tokens, but Anthropic's
+            # input_tokens EXCLUDES cached tokens. We extract cached tokens and subtract.
+            if "usage" in chunk and chunk["usage"]:
+                usage = chunk["usage"]
+                input_tokens = usage.get("prompt_tokens", input_tokens)
+                output_tokens = usage.get("completion_tokens", output_tokens)
+                # Extract cached tokens from prompt_tokens_details
+                if usage.get("prompt_tokens_details"):
+                    cached_tokens = usage["prompt_tokens_details"].get(
+                        "cached_tokens", cached_tokens
+                    )
+
+            # Send message_start on first chunk
+            if not message_started:
+                # Build usage with cached tokens properly handled for Anthropic format
+                usage_dict = {
+                    "input_tokens": input_tokens - cached_tokens,
+                    "output_tokens": 0,
+                }
+                if cached_tokens > 0:
+                    usage_dict["cache_read_input_tokens"] = cached_tokens
+                    usage_dict["cache_creation_input_tokens"] = 0
+
+                message_start = {
+                    "type": "message_start",
+                    "message": {
+                        "id": request_id,
+                        "type": "message",
+                        "role": "assistant",
+                        "content": [],
+                        "model": original_model,
+                        "stop_reason": None,
+                        "stop_sequence": None,
+                        "usage": usage_dict,
+                    },
+                }
+                yield f"event: message_start\ndata: {json.dumps(message_start)}\n\n"
+                message_started = True
+
+            choices = chunk.get("choices") or []
+            if not choices:
+                continue
+
+            delta = choices[0].get("delta", {})
+
+            # Handle reasoning/thinking content (from OpenAI-style reasoning_content)
+            reasoning_content = delta.get("reasoning_content")
+            if reasoning_content:
+                if not thinking_block_started:
+                    # Start a thinking content block
+                    block_start = {
+                        "type": "content_block_start",
+                        "index": current_block_index,
+                        "content_block": {"type": "thinking", "thinking": ""},
+                    }
+                    yield f"event: content_block_start\ndata: {json.dumps(block_start)}\n\n"
+                    thinking_block_started = True
+
+                # Send thinking delta
+                block_delta = {
+                    "type": "content_block_delta",
+                    "index": current_block_index,
+                    "delta": {"type": "thinking_delta", "thinking": reasoning_content},
+                }
+                yield f"event: content_block_delta\ndata: {json.dumps(block_delta)}\n\n"
+                # Accumulate thinking for logging
+                accumulated_thinking += reasoning_content
+
+            # Handle text content
+            content = delta.get("content")
+            if content:
+                # If we were in a thinking block, close it first
+                if thinking_block_started and not content_block_started:
+                    yield f'event: content_block_stop\ndata: {{"type": "content_block_stop", "index": {current_block_index}}}\n\n'
+                    current_block_index += 1
+                    thinking_block_started = False
+
+                if not content_block_started:
+                    # Start a text content block
+                    block_start = {
+                        "type": "content_block_start",
+                        "index": current_block_index,
+                        "content_block": {"type": "text", "text": ""},
+                    }
+                    yield f"event: content_block_start\ndata: {json.dumps(block_start)}\n\n"
+                    content_block_started = True
+
+                # Send content delta
+                block_delta = {
+                    "type": "content_block_delta",
+                    "index": current_block_index,
+                    "delta": {"type": "text_delta", "text": content},
+                }
+                yield f"event: content_block_delta\ndata: {json.dumps(block_delta)}\n\n"
+                # Accumulate text for logging
+                accumulated_text += content
+
+            # Handle tool calls
+            # Use `or []` to handle providers that send "tool_calls": null
+            tool_calls = delta.get("tool_calls") or []
+            for tc in tool_calls:
+                tc_index = tc.get("index", 0)
+
+                if tc_index not in tool_calls_by_index:
+                    # Close previous thinking block if open
+                    if thinking_block_started:
+                        yield f'event: content_block_stop\ndata: {{"type": "content_block_stop", "index": {current_block_index}}}\n\n'
+                        current_block_index += 1
+                        thinking_block_started = False
+
+                    # Close previous text block if open
+                    if content_block_started:
+                        yield f'event: content_block_stop\ndata: {{"type": "content_block_stop", "index": {current_block_index}}}\n\n'
+                        current_block_index += 1
+                        content_block_started = False
+
+                    # Start new tool use block
+                    tool_calls_by_index[tc_index] = {
+                        "id": tc.get("id", f"toolu_{uuid.uuid4().hex[:12]}"),
+                        "name": tc.get("function", {}).get("name", ""),
+                        "arguments": "",
+                    }
+                    # Track which block index this tool call uses
+                    tool_block_indices[tc_index] = current_block_index
+
+                    block_start = {
+                        "type": "content_block_start",
+                        "index": current_block_index,
+                        "content_block": {
+                            "type": "tool_use",
+                            "id": tool_calls_by_index[tc_index]["id"],
+                            "name": tool_calls_by_index[tc_index]["name"],
+                            "input": {},
+                        },
+                    }
+                    yield f"event: content_block_start\ndata: {json.dumps(block_start)}\n\n"
+                    # Increment for the next block
+                    current_block_index += 1
+
+                # Accumulate arguments
+                func = tc.get("function", {})
+                if func.get("name"):
+                    tool_calls_by_index[tc_index]["name"] = func["name"]
+                if func.get("arguments"):
+                    tool_calls_by_index[tc_index]["arguments"] += func["arguments"]
+
+                    # Send partial JSON delta using the correct block index for this tool
+                    block_delta = {
+                        "type": "content_block_delta",
+                        "index": tool_block_indices[tc_index],
+                        "delta": {
+                            "type": "input_json_delta",
+                            "partial_json": func["arguments"],
+                        },
+                    }
+                    yield f"event: content_block_delta\ndata: {json.dumps(block_delta)}\n\n"
+
+            # Note: We intentionally ignore finish_reason here.
+            # Block closing is handled when we receive [DONE] to avoid
+            # premature closes with providers that send finish_reason on each chunk.
+
+    except Exception as e:
+        logger.error(f"Error in Anthropic streaming wrapper: {e}")
+
+        # If we haven't sent message_start yet, send it now so the client can display the error
+        # Claude Code and other clients may ignore events that come before message_start
+        if not message_started:
+            # Build usage with cached tokens properly handled
+            usage_dict = {
+                "input_tokens": input_tokens - cached_tokens,
+                "output_tokens": 0,
+            }
+            if cached_tokens > 0:
+                usage_dict["cache_read_input_tokens"] = cached_tokens
+                usage_dict["cache_creation_input_tokens"] = 0
+
+            message_start = {
+                "type": "message_start",
+                "message": {
+                    "id": request_id,
+                    "type": "message",
+                    "role": "assistant",
+                    "content": [],
+                    "model": original_model,
+                    "stop_reason": None,
+                    "stop_sequence": None,
+                    "usage": usage_dict,
+                },
+            }
+            yield f"event: message_start\ndata: {json.dumps(message_start)}\n\n"
+
+        # Send the error as a text content block so it's visible to the user
+        error_message = f"Error: {str(e)}"
+        error_block_start = {
+            "type": "content_block_start",
+            "index": current_block_index,
+            "content_block": {"type": "text", "text": ""},
+        }
+        yield f"event: content_block_start\ndata: {json.dumps(error_block_start)}\n\n"
+
+        error_block_delta = {
+            "type": "content_block_delta",
+            "index": current_block_index,
+            "delta": {"type": "text_delta", "text": error_message},
+        }
+        yield f"event: content_block_delta\ndata: {json.dumps(error_block_delta)}\n\n"
+
+        yield f'event: content_block_stop\ndata: {{"type": "content_block_stop", "index": {current_block_index}}}\n\n'
+
+        # Build final usage with cached tokens
+        final_usage = {"output_tokens": 0}
+        if cached_tokens > 0:
+            final_usage["cache_read_input_tokens"] = cached_tokens
+            final_usage["cache_creation_input_tokens"] = 0
+
+        # Send message_delta and message_stop to properly close the stream
+        yield f'event: message_delta\ndata: {{"type": "message_delta", "delta": {{"stop_reason": "end_turn", "stop_sequence": null}}, "usage": {json.dumps(final_usage)}}}\n\n'
+        yield 'event: message_stop\ndata: {"type": "message_stop"}\n\n'
+
+        # Also send the formal error event for clients that handle it
+        error_event = {
+            "type": "error",
+            "error": {"type": "api_error", "message": str(e)},
+        }
+        yield f"event: error\ndata: {json.dumps(error_event)}\n\n"

src/rotator_library/anthropic_compat/translator.py

@@ -0,0 +1,629 @@
+# SPDX-License-Identifier: LGPL-3.0-only
+# Copyright (c) 2026 Mirrowel
+
+"""
+Format translation functions between Anthropic and OpenAI API formats.
+
+This module provides functions to convert requests and responses between
+Anthropic's Messages API format and OpenAI's Chat Completions API format.
+This enables any OpenAI-compatible provider to work with Anthropic clients.
+"""
+
+import json
+import uuid
+from typing import Any, Dict, List, Optional, Union
+
+from .models import AnthropicMessagesRequest
+
+MIN_THINKING_SIGNATURE_LENGTH = 100
+
+# =============================================================================
+# THINKING BUDGET TO REASONING EFFORT MAPPING
+# =============================================================================
+
+# Budget thresholds for reasoning effort levels (based on token counts)
+# These map Anthropic's budget_tokens to OpenAI-style reasoning_effort levels
+THINKING_BUDGET_THRESHOLDS = {
+    "minimal": 4096,
+    "low": 8192,
+    "low_medium": 12288,
+    "medium": 16384,
+    "medium_high": 24576,
+    "high": 32768,
+}
+
+# Providers that support granular reasoning effort levels (low_medium, medium_high, etc.)
+# Other providers will receive simplified levels (low, medium, high)
+GRANULAR_REASONING_PROVIDERS = {"antigravity"}
+
+
+def _budget_to_reasoning_effort(budget_tokens: int, model: str) -> str:
+    """
+    Map Anthropic thinking budget_tokens to a reasoning_effort level.
+
+    Args:
+        budget_tokens: The thinking budget in tokens from the Anthropic request
+        model: The model name (used to determine if provider supports granular levels)
+
+    Returns:
+        A reasoning_effort level string (e.g., "low", "medium", "high")
+    """
+    # Determine granular level based on budget
+    if budget_tokens <= THINKING_BUDGET_THRESHOLDS["minimal"]:
+        granular_level = "minimal"
+    elif budget_tokens <= THINKING_BUDGET_THRESHOLDS["low"]:
+        granular_level = "low"
+    elif budget_tokens <= THINKING_BUDGET_THRESHOLDS["low_medium"]:
+        granular_level = "low_medium"
+    elif budget_tokens <= THINKING_BUDGET_THRESHOLDS["medium"]:
+        granular_level = "medium"
+    elif budget_tokens <= THINKING_BUDGET_THRESHOLDS["medium_high"]:
+        granular_level = "medium_high"
+    else:
+        granular_level = "high"
+
+    # Check if provider supports granular levels
+    provider = model.split("/")[0].lower() if "/" in model else ""
+    if provider in GRANULAR_REASONING_PROVIDERS:
+        return granular_level
+
+    # Simplify to basic levels for non-granular providers
+    simplify_map = {
+        "minimal": "low",
+        "low": "low",
+        "low_medium": "medium",
+        "medium": "medium",
+        "medium_high": "high",
+        "high": "high",
+    }
+    return simplify_map.get(granular_level, "medium")
+
+
+def _reorder_assistant_content(content: List[dict]) -> List[dict]:
+    """
+    Reorder assistant message content blocks to ensure correct order:
+    1. Thinking blocks come first (required when thinking is enabled)
+    2. Text blocks come in the middle (filtering out empty ones)
+    3. Tool_use blocks come at the end (required before tool_result)
+
+    This matches Anthropic's expected ordering and prevents API errors.
+    """
+    if not isinstance(content, list) or len(content) <= 1:
+        return content
+
+    thinking_blocks = []
+    text_blocks = []
+    tool_use_blocks = []
+    other_blocks = []
+
+    for block in content:
+        if not isinstance(block, dict):
+            other_blocks.append(block)
+            continue
+
+        block_type = block.get("type", "")
+
+        if block_type in ("thinking", "redacted_thinking"):
+            # Sanitize thinking blocks - remove cache_control and other extra fields
+            sanitized = {
+                "type": block_type,
+                "thinking": block.get("thinking", ""),
+            }
+            if block.get("signature"):
+                sanitized["signature"] = block["signature"]
+            thinking_blocks.append(sanitized)
+
+        elif block_type == "tool_use":
+            tool_use_blocks.append(block)
+
+        elif block_type == "text":
+            # Only keep text blocks with meaningful content
+            text = block.get("text", "")
+            if text and text.strip():
+                text_blocks.append(block)
+
+        else:
+            # Other block types (images, documents, etc.) go in the text position
+            other_blocks.append(block)
+
+    # Reorder: thinking → other → text → tool_use
+    return thinking_blocks + other_blocks + text_blocks + tool_use_blocks
+
+
+def anthropic_to_openai_messages(
+    anthropic_messages: List[dict], system: Optional[Union[str, List[dict]]] = None
+) -> List[dict]:
+    """
+    Convert Anthropic message format to OpenAI format.
+
+    Key differences:
+    - Anthropic: system is a separate field, content can be string or list of blocks
+    - OpenAI: system is a message with role="system", content is usually string
+
+    Args:
+        anthropic_messages: List of messages in Anthropic format
+        system: Optional system message (string or list of text blocks)
+
+    Returns:
+        List of messages in OpenAI format
+    """
+    openai_messages = []
+
+    # Handle system message
+    if system:
+        if isinstance(system, str):
+            openai_messages.append({"role": "system", "content": system})
+        elif isinstance(system, list):
+            # System can be list of text blocks in Anthropic format
+            system_text = " ".join(
+                block.get("text", "")
+                for block in system
+                if isinstance(block, dict) and block.get("type") == "text"
+            )
+            if system_text:
+                openai_messages.append({"role": "system", "content": system_text})
+
+    for msg in anthropic_messages:
+        role = msg.get("role", "user")
+        content = msg.get("content", "")
+
+        if isinstance(content, str):
+            openai_messages.append({"role": role, "content": content})
+        elif isinstance(content, list):
+            # Reorder assistant content blocks to ensure correct order:
+            # thinking → text → tool_use
+            if role == "assistant":
+                content = _reorder_assistant_content(content)
+
+            # Handle content blocks
+            openai_content = []
+            tool_calls = []
+            reasoning_content = ""
+            thinking_signature = ""
+
+            for block in content:
+                if isinstance(block, dict):
+                    block_type = block.get("type", "text")
+
+                    if block_type == "text":
+                        openai_content.append(
+                            {"type": "text", "text": block.get("text", "")}
+                        )
+                    elif block_type == "image":
+                        # Convert Anthropic image format to OpenAI
+                        source = block.get("source", {})
+                        if source.get("type") == "base64":
+                            openai_content.append(
+                                {
+                                    "type": "image_url",
+                                    "image_url": {
+                                        "url": f"data:{source.get('media_type', 'image/png')};base64,{source.get('data', '')}"
+                                    },
+                                }
+                            )
+                        elif source.get("type") == "url":
+                            openai_content.append(
+                                {
+                                    "type": "image_url",
+                                    "image_url": {"url": source.get("url", "")},
+                                }
+                            )
+                    elif block_type == "document":
+                        # Convert Anthropic document format (e.g. PDF) to OpenAI
+                        # Documents are treated similarly to images with appropriate mime type
+                        source = block.get("source", {})
+                        if source.get("type") == "base64":
+                            openai_content.append(
+                                {
+                                    "type": "image_url",
+                                    "image_url": {
+                                        "url": f"data:{source.get('media_type', 'application/pdf')};base64,{source.get('data', '')}"
+                                    },
+                                }
+                            )
+                        elif source.get("type") == "url":
+                            openai_content.append(
+                                {
+                                    "type": "image_url",
+                                    "image_url": {"url": source.get("url", "")},
+                                }
+                            )
+                    elif block_type == "thinking":
+                        signature = block.get("signature", "")
+                        if (
+                            signature
+                            and len(signature) >= MIN_THINKING_SIGNATURE_LENGTH
+                        ):
+                            thinking_text = block.get("thinking", "")
+                            if thinking_text:
+                                reasoning_content += thinking_text
+                            thinking_signature = signature
+                    elif block_type == "redacted_thinking":
+                        signature = block.get("signature", "")
+                        if (
+                            signature
+                            and len(signature) >= MIN_THINKING_SIGNATURE_LENGTH
+                        ):
+                            thinking_signature = signature
+                    elif block_type == "tool_use":
+                        # Anthropic tool_use -> OpenAI tool_calls
+                        tool_calls.append(
+                            {
+                                "id": block.get("id", ""),
+                                "type": "function",
+                                "function": {
+                                    "name": block.get("name", ""),
+                                    "arguments": json.dumps(block.get("input", {})),
+                                },
+                            }
+                        )
+                    elif block_type == "tool_result":
+                        # Tool results become separate messages in OpenAI format
+                        # Content can be string, or list of text/image blocks
+                        tool_content = block.get("content", "")
+                        if isinstance(tool_content, str):
+                            # Simple string content
+                            openai_messages.append(
+                                {
+                                    "role": "tool",
+                                    "tool_call_id": block.get("tool_use_id", ""),
+                                    "content": tool_content,
+                                }
+                            )
+                        elif isinstance(tool_content, list):
+                            # List of content blocks - may include text and images
+                            tool_content_parts = []
+                            for b in tool_content:
+                                if not isinstance(b, dict):
+                                    continue
+                                b_type = b.get("type", "")
+                                if b_type == "text":
+                                    tool_content_parts.append(
+                                        {"type": "text", "text": b.get("text", "")}
+                                    )
+                                elif b_type == "image":
+                                    # Convert Anthropic image format to OpenAI format
+                                    source = b.get("source", {})
+                                    if source.get("type") == "base64":
+                                        tool_content_parts.append(
+                                            {
+                                                "type": "image_url",
+                                                "image_url": {
+                                                    "url": f"data:{source.get('media_type', 'image/png')};base64,{source.get('data', '')}"
+                                                },
+                                            }
+                                        )
+                                    elif source.get("type") == "url":
+                                        tool_content_parts.append(
+                                            {
+                                                "type": "image_url",
+                                                "image_url": {
+                                                    "url": source.get("url", "")
+                                                },
+                                            }
+                                        )
+
+                            # If we only have text parts, join them as a string for compatibility
+                            # Otherwise use the array format for multimodal content
+                            if all(p.get("type") == "text" for p in tool_content_parts):
+                                combined_text = " ".join(
+                                    p.get("text", "") for p in tool_content_parts
+                                )
+                                openai_messages.append(
+                                    {
+                                        "role": "tool",
+                                        "tool_call_id": block.get("tool_use_id", ""),
+                                        "content": combined_text,
+                                    }
+                                )
+                            elif tool_content_parts:
+                                # Multimodal content (includes images)
+                                openai_messages.append(
+                                    {
+                                        "role": "tool",
+                                        "tool_call_id": block.get("tool_use_id", ""),
+                                        "content": tool_content_parts,
+                                    }
+                                )
+                            else:
+                                # Empty content
+                                openai_messages.append(
+                                    {
+                                        "role": "tool",
+                                        "tool_call_id": block.get("tool_use_id", ""),
+                                        "content": "",
+                                    }
+                                )
+                        else:
+                            # Fallback for unexpected content type
+                            openai_messages.append(
+                                {
+                                    "role": "tool",
+                                    "tool_call_id": block.get("tool_use_id", ""),
+                                    "content": str(tool_content)
+                                    if tool_content
+                                    else "",
+                                }
+                            )
+                        continue  # Don't add to current message
+
+            # Build the message
+            if tool_calls:
+                # Assistant message with tool calls
+                msg_dict = {"role": role}
+                if openai_content:
+                    # If there's text content alongside tool calls
+                    text_parts = [
+                        c.get("text", "")
+                        for c in openai_content
+                        if c.get("type") == "text"
+                    ]
+                    msg_dict["content"] = " ".join(text_parts) if text_parts else None
+                else:
+                    msg_dict["content"] = None
+                if reasoning_content:
+                    msg_dict["reasoning_content"] = reasoning_content
+                if thinking_signature:
+                    msg_dict["thinking_signature"] = thinking_signature
+                msg_dict["tool_calls"] = tool_calls
+                openai_messages.append(msg_dict)
+            elif openai_content:
+                # Check if it's just text or mixed content
+                if len(openai_content) == 1 and openai_content[0].get("type") == "text":
+                    msg_dict = {
+                        "role": role,
+                        "content": openai_content[0].get("text", ""),
+                    }
+                    if reasoning_content:
+                        msg_dict["reasoning_content"] = reasoning_content
+                    if thinking_signature:
+                        msg_dict["thinking_signature"] = thinking_signature
+                    openai_messages.append(msg_dict)
+                else:
+                    msg_dict = {"role": role, "content": openai_content}
+                    if reasoning_content:
+                        msg_dict["reasoning_content"] = reasoning_content
+                    if thinking_signature:
+                        msg_dict["thinking_signature"] = thinking_signature
+                    openai_messages.append(msg_dict)
+            elif reasoning_content:
+                msg_dict = {"role": role, "content": ""}
+                msg_dict["reasoning_content"] = reasoning_content
+                if thinking_signature:
+                    msg_dict["thinking_signature"] = thinking_signature
+                openai_messages.append(msg_dict)
+
+    return openai_messages
+
+
+def anthropic_to_openai_tools(
+    anthropic_tools: Optional[List[dict]],
+) -> Optional[List[dict]]:
+    """
+    Convert Anthropic tool definitions to OpenAI format.
+
+    Args:
+        anthropic_tools: List of tools in Anthropic format
+
+    Returns:
+        List of tools in OpenAI format, or None if no tools provided
+    """
+    if not anthropic_tools:
+        return None
+
+    openai_tools = []
+    for tool in anthropic_tools:
+        openai_tools.append(
+            {
+                "type": "function",
+                "function": {
+                    "name": tool.get("name", ""),
+                    "description": tool.get("description", ""),
+                    "parameters": tool.get("input_schema", {}),
+                },
+            }
+        )
+    return openai_tools
+
+
+def anthropic_to_openai_tool_choice(
+    anthropic_tool_choice: Optional[dict],
+) -> Optional[Union[str, dict]]:
+    """
+    Convert Anthropic tool_choice to OpenAI format.
+
+    Args:
+        anthropic_tool_choice: Tool choice in Anthropic format
+
+    Returns:
+        Tool choice in OpenAI format
+    """
+    if not anthropic_tool_choice:
+        return None
+
+    choice_type = anthropic_tool_choice.get("type", "auto")
+
+    if choice_type == "auto":
+        return "auto"
+    elif choice_type == "any":
+        return "required"
+    elif choice_type == "tool":
+        return {
+            "type": "function",
+            "function": {"name": anthropic_tool_choice.get("name", "")},
+        }
+    elif choice_type == "none":
+        return "none"
+
+    return "auto"
+
+
+def openai_to_anthropic_response(openai_response: dict, original_model: str) -> dict:
+    """
+    Convert OpenAI chat completion response to Anthropic Messages format.
+
+    Args:
+        openai_response: Response from OpenAI-compatible API
+        original_model: The model name requested by the client
+
+    Returns:
+        Response in Anthropic Messages format
+    """
+    choice = openai_response.get("choices", [{}])[0]
+    message = choice.get("message", {})
+    usage = openai_response.get("usage", {})
+
+    # Build content blocks
+    content_blocks = []
+
+    # Add thinking content block if reasoning_content is present
+    reasoning_content = message.get("reasoning_content")
+    if reasoning_content:
+        thinking_signature = message.get("thinking_signature", "")
+        signature = (
+            thinking_signature
+            if thinking_signature
+            and len(thinking_signature) >= MIN_THINKING_SIGNATURE_LENGTH
+            else ""
+        )
+        content_blocks.append(
+            {
+                "type": "thinking",
+                "thinking": reasoning_content,
+                "signature": signature,
+            }
+        )
+
+    # Add text content if present
+    text_content = message.get("content")
+    if text_content:
+        content_blocks.append({"type": "text", "text": text_content})
+
+    # Add tool use blocks if present
+    tool_calls = message.get("tool_calls") or []
+    for tc in tool_calls:
+        func = tc.get("function", {})
+        try:
+            input_data = json.loads(func.get("arguments", "{}"))
+        except json.JSONDecodeError:
+            input_data = {}
+
+        content_blocks.append(
+            {
+                "type": "tool_use",
+                "id": tc.get("id", f"toolu_{uuid.uuid4().hex[:12]}"),
+                "name": func.get("name", ""),
+                "input": input_data,
+            }
+        )
+
+    # Map finish_reason to stop_reason
+    finish_reason = choice.get("finish_reason", "end_turn")
+    stop_reason_map = {
+        "stop": "end_turn",
+        "length": "max_tokens",
+        "tool_calls": "tool_use",
+        "content_filter": "end_turn",
+        "function_call": "tool_use",
+    }
+    stop_reason = stop_reason_map.get(finish_reason, "end_turn")
+
+    # Build usage
+    # Note: Google's promptTokenCount INCLUDES cached tokens, but Anthropic's
+    # input_tokens EXCLUDES cached tokens. We need to subtract cached tokens.
+    prompt_tokens = usage.get("prompt_tokens", 0)
+    cached_tokens = 0
+
+    # Extract cached tokens if present
+    if usage.get("prompt_tokens_details"):
+        details = usage["prompt_tokens_details"]
+        cached_tokens = details.get("cached_tokens", 0)
+
+    anthropic_usage = {
+        "input_tokens": prompt_tokens - cached_tokens,  # Subtract cached tokens
+        "output_tokens": usage.get("completion_tokens", 0),
+    }
+
+    # Add cache tokens if present
+    if cached_tokens > 0:
+        anthropic_usage["cache_read_input_tokens"] = cached_tokens
+        anthropic_usage["cache_creation_input_tokens"] = 0
+
+    return {
+        "id": openai_response.get("id", f"msg_{uuid.uuid4().hex[:24]}"),
+        "type": "message",
+        "role": "assistant",
+        "content": content_blocks,
+        "model": original_model,
+        "stop_reason": stop_reason,
+        "stop_sequence": None,
+        "usage": anthropic_usage,
+    }
+
+
+def translate_anthropic_request(request: AnthropicMessagesRequest) -> Dict[str, Any]:
+    """
+    Translate a complete Anthropic Messages API request to OpenAI format.
+
+    This is a high-level function that handles all aspects of request translation,
+    including messages, tools, tool_choice, and thinking configuration.
+
+    Args:
+        request: An AnthropicMessagesRequest object
+
+    Returns:
+        Dictionary containing the OpenAI-compatible request parameters
+    """
+    anthropic_request = request.model_dump(exclude_none=True)
+
+    messages = anthropic_request.get("messages", [])
+    openai_messages = anthropic_to_openai_messages(
+        messages, anthropic_request.get("system")
+    )
+
+    openai_tools = anthropic_to_openai_tools(anthropic_request.get("tools"))
+    openai_tool_choice = anthropic_to_openai_tool_choice(
+        anthropic_request.get("tool_choice")
+    )
+
+    # Build OpenAI-compatible request
+    openai_request = {
+        "model": request.model,
+        "messages": openai_messages,
+        "max_tokens": request.max_tokens,
+        "stream": request.stream or False,
+    }
+
+    if request.temperature is not None:
+        openai_request["temperature"] = request.temperature
+    if request.top_p is not None:
+        openai_request["top_p"] = request.top_p
+    if request.top_k is not None:
+        openai_request["top_k"] = request.top_k
+    if request.stop_sequences:
+        openai_request["stop"] = request.stop_sequences
+    if openai_tools:
+        openai_request["tools"] = openai_tools
+    if openai_tool_choice:
+        openai_request["tool_choice"] = openai_tool_choice
+
+    # Note: request.metadata is intentionally not mapped.
+    # OpenAI's API doesn't have an equivalent field for client-side metadata.
+    # The metadata is typically used by Anthropic clients for tracking purposes
+    # and doesn't affect the model's behavior.
+
+    # Handle Anthropic thinking config -> reasoning_effort translation
+    # Only set reasoning_effort if thinking is explicitly configured
+    if request.thinking:
+        if request.thinking.type == "enabled":
+            # Only set reasoning_effort if budget_tokens was specified
+            if request.thinking.budget_tokens is not None:
+                openai_request["reasoning_effort"] = _budget_to_reasoning_effort(
+                    request.thinking.budget_tokens, request.model
+                )
+            # If thinking enabled but no budget specified, don't set anything
+            # Let the provider decide the default
+        elif request.thinking.type == "disabled":
+            openai_request["reasoning_effort"] = "disable"
+
+    return openai_request

src/rotator_library/background_refresher.py

@@ -0,0 +1,289 @@
+# SPDX-License-Identifier: LGPL-3.0-only
+# Copyright (c) 2026 Mirrowel
+
+# src/rotator_library/background_refresher.py
+
+import os
+import asyncio
+import logging
+from typing import TYPE_CHECKING, Optional, Dict, Any, List
+
+if TYPE_CHECKING:
+    from .client import RotatingClient
+
+lib_logger = logging.getLogger("rotator_library")
+
+# =============================================================================
+# CONFIGURATION DEFAULTS
+# =============================================================================
+# These can be overridden via environment variables.
+
+# OAuth token refresh interval in seconds
+# Override: OAUTH_REFRESH_INTERVAL=<seconds>
+DEFAULT_OAUTH_REFRESH_INTERVAL: int = 600  # 10 minutes
+
+# Default interval for provider background jobs (quota refresh, etc.)
+# Individual providers can override this in their get_background_job_config()
+DEFAULT_BACKGROUND_JOB_INTERVAL: int = 300  # 5 minutes
+
+# Whether to run background jobs immediately on start (before first interval)
+DEFAULT_BACKGROUND_JOB_RUN_ON_START: bool = True
+
+
+class BackgroundRefresher:
+    """
+    A background task manager that handles:
+    1. Periodic OAuth token refresh for all providers
+    2. Provider-specific background jobs (e.g., quota refresh) with independent timers
+
+    Each provider can define its own background job via get_background_job_config()
+    and run_background_job(). These run on their own schedules, independent of the
+    OAuth refresh interval.
+    """
+
+    def __init__(self, client: "RotatingClient"):
+        self._client = client
+        self._task: Optional[asyncio.Task] = None
+        self._provider_job_tasks: Dict[str, asyncio.Task] = {}  # provider -> task
+        self._initialized = False
+        try:
+            interval_str = os.getenv(
+                "OAUTH_REFRESH_INTERVAL", str(DEFAULT_OAUTH_REFRESH_INTERVAL)
+            )
+            self._interval = int(interval_str)
+        except ValueError:
+            lib_logger.warning(
+                f"Invalid OAUTH_REFRESH_INTERVAL '{interval_str}'. "
+                f"Falling back to {DEFAULT_OAUTH_REFRESH_INTERVAL}s."
+            )
+            self._interval = DEFAULT_OAUTH_REFRESH_INTERVAL
+
+    def start(self):
+        """Starts the background refresh task."""
+        if self._task is None:
+            self._task = asyncio.create_task(self._run())
+            lib_logger.info(
+                f"Background token refresher started. Check interval: {self._interval} seconds."
+            )
+
+    async def stop(self):
+        """Stops all background tasks (main loop + provider jobs)."""
+        # Cancel provider job tasks first
+        for provider, task in self._provider_job_tasks.items():
+            if task and not task.done():
+                task.cancel()
+                try:
+                    await task
+                except asyncio.CancelledError:
+                    pass
+                lib_logger.debug(f"Stopped background job for '{provider}'")
+
+        self._provider_job_tasks.clear()
+
+        # Cancel main task
+        if self._task:
+            self._task.cancel()
+            try:
+                await self._task
+            except asyncio.CancelledError:
+                pass
+            lib_logger.info("Background token refresher stopped.")
+
+    async def _initialize_credentials(self):
+        """
+        Initialize all providers by loading credentials and persisted tier data.
+        Called once before the main refresh loop starts.
+        """
+        if self._initialized:
+            return
+
+        api_summary = {}  # provider -> count
+        oauth_summary = {}  # provider -> {"count": N, "tiers": {tier: count}}
+
+        all_credentials = self._client.all_credentials
+        oauth_providers = self._client.oauth_providers
+
+        for provider, credentials in all_credentials.items():
+            if not credentials:
+                continue
+
+            provider_plugin = self._client._get_provider_instance(provider)
+
+            # Call initialize_credentials if provider supports it
+            if provider_plugin and hasattr(provider_plugin, "initialize_credentials"):
+                try:
+                    await provider_plugin.initialize_credentials(credentials)
+                except Exception as e:
+                    lib_logger.error(
+                        f"Error initializing credentials for provider '{provider}': {e}"
+                    )
+
+            # Build summary based on provider type
+            if provider in oauth_providers:
+                tier_breakdown = {}
+                if provider_plugin and hasattr(
+                    provider_plugin, "get_credential_tier_name"
+                ):
+                    for cred in credentials:
+                        tier = provider_plugin.get_credential_tier_name(cred)
+                        if tier:
+                            tier_breakdown[tier] = tier_breakdown.get(tier, 0) + 1
+                oauth_summary[provider] = {
+                    "count": len(credentials),
+                    "tiers": tier_breakdown,
+                }
+            else:
+                api_summary[provider] = len(credentials)
+
+        # Log 3-line summary
+        total_providers = len(api_summary) + len(oauth_summary)
+        total_credentials = sum(api_summary.values()) + sum(
+            d["count"] for d in oauth_summary.values()
+        )
+
+        if total_providers > 0:
+            lib_logger.info(
+                f"Providers initialized: {total_providers} providers, {total_credentials} credentials"
+            )
+
+            # API providers line
+            if api_summary:
+                api_parts = [f"{p}:{c}" for p, c in sorted(api_summary.items())]
+                lib_logger.info(f"  API: {', '.join(api_parts)}")
+
+            # OAuth providers line with tier breakdown
+            if oauth_summary:
+                oauth_parts = []
+                for provider, data in sorted(oauth_summary.items()):
+                    if data["tiers"]:
+                        tier_str = ", ".join(
+                            f"{t}:{c}" for t, c in sorted(data["tiers"].items())
+                        )
+                        oauth_parts.append(f"{provider}:{data['count']} ({tier_str})")
+                    else:
+                        oauth_parts.append(f"{provider}:{data['count']}")
+                lib_logger.info(f"  OAuth: {', '.join(oauth_parts)}")
+
+        self._initialized = True
+
+    def _start_provider_background_jobs(self):
+        """
+        Start independent background job tasks for providers that define them.
+
+        Each provider with a get_background_job_config() that returns a config
+        gets its own asyncio task running on its own schedule.
+        """
+        all_credentials = self._client.all_credentials
+
+        for provider, credentials in all_credentials.items():
+            if not credentials:
+                lib_logger.debug(f"Skipping {provider} background job: no credentials")
+                continue
+
+            provider_plugin = self._client._get_provider_instance(provider)
+            if not provider_plugin:
+                lib_logger.debug(
+                    f"Skipping {provider} background job: no provider instance"
+                )
+                continue
+
+            # Check if provider has a background job
+            if not hasattr(provider_plugin, "get_background_job_config"):
+                lib_logger.debug(
+                    f"Skipping {provider} background job: no get_background_job_config method"
+                )
+                continue
+
+            config = provider_plugin.get_background_job_config()
+            if not config:
+                lib_logger.debug(f"Skipping {provider} background job: config is None")
+                continue
+
+            # Start the provider's background job task
+            task = asyncio.create_task(
+                self._run_provider_background_job(
+                    provider, provider_plugin, credentials, config
+                )
+            )
+            self._provider_job_tasks[provider] = task
+
+            job_name = config.get("name", "background_job")
+            interval = config.get("interval", DEFAULT_BACKGROUND_JOB_INTERVAL)
+            lib_logger.info(f"Started {provider} {job_name} (interval: {interval}s)")
+
+    async def _run_provider_background_job(
+        self,
+        provider_name: str,
+        provider: Any,
+        credentials: List[str],
+        config: Dict[str, Any],
+    ) -> None:
+        """
+        Independent loop for a single provider's background job.
+
+        Args:
+            provider_name: Name of the provider (for logging)
+            provider: Provider plugin instance
+            credentials: List of credential paths for this provider
+            config: Background job configuration from get_background_job_config()
+        """
+        interval = config.get("interval", DEFAULT_BACKGROUND_JOB_INTERVAL)
+        job_name = config.get("name", "background_job")
+        run_on_start = config.get("run_on_start", DEFAULT_BACKGROUND_JOB_RUN_ON_START)
+
+        # Run immediately on start if configured
+        if run_on_start:
+            try:
+                await provider.run_background_job(
+                    self._client.usage_manager, credentials
+                )
+                lib_logger.debug(f"{provider_name} {job_name}: initial run complete")
+            except Exception as e:
+                lib_logger.error(
+                    f"Error in {provider_name} {job_name} (initial run): {e}"
+                )
+
+        # Main loop
+        while True:
+            try:
+                await asyncio.sleep(interval)
+                await provider.run_background_job(
+                    self._client.usage_manager, credentials
+                )
+                lib_logger.debug(f"{provider_name} {job_name}: periodic run complete")
+            except asyncio.CancelledError:
+                lib_logger.debug(f"{provider_name} {job_name}: cancelled")
+                break
+            except Exception as e:
+                lib_logger.error(f"Error in {provider_name} {job_name}: {e}")
+
+    async def _run(self):
+        """The main loop for OAuth token refresh."""
+        # Initialize credentials (load persisted tiers) before starting
+        await self._initialize_credentials()
+
+        # Start provider-specific background jobs with their own timers
+        self._start_provider_background_jobs()
+
+        # Main OAuth refresh loop
+        while True:
+            try:
+                oauth_configs = self._client.get_oauth_credentials()
+                for provider, paths in oauth_configs.items():
+                    provider_plugin = self._client._get_provider_instance(provider)
+                    if provider_plugin and hasattr(
+                        provider_plugin, "proactively_refresh"
+                    ):
+                        for path in paths:
+                            try:
+                                await provider_plugin.proactively_refresh(path)
+                            except Exception as e:
+                                lib_logger.error(
+                                    f"Error during proactive refresh for '{path}': {e}"
+                                )
+
+                await asyncio.sleep(self._interval)
+            except asyncio.CancelledError:
+                break
+            except Exception as e:
+                lib_logger.error(f"Unexpected error in background refresher loop: {e}")

src/rotator_library/config/__init__.py

@@ -0,0 +1,60 @@
+# SPDX-License-Identifier: LGPL-3.0-only
+# Copyright (c) 2026 Mirrowel
+
+"""
+Configuration module for the rotator library.
+
+Exports all centralized defaults for use across the library.
+"""
+
+from .defaults import (
+    # Rotation & Selection
+    DEFAULT_ROTATION_MODE,
+    DEFAULT_ROTATION_TOLERANCE,
+    DEFAULT_MAX_RETRIES,
+    DEFAULT_GLOBAL_TIMEOUT,
+    # Tier & Priority
+    DEFAULT_TIER_PRIORITY,
+    DEFAULT_SEQUENTIAL_FALLBACK_MULTIPLIER,
+    # Fair Cycle Rotation
+    DEFAULT_FAIR_CYCLE_ENABLED,
+    DEFAULT_FAIR_CYCLE_TRACKING_MODE,
+    DEFAULT_FAIR_CYCLE_CROSS_TIER,
+    DEFAULT_FAIR_CYCLE_DURATION,
+    DEFAULT_EXHAUSTION_COOLDOWN_THRESHOLD,
+    # Custom Caps
+    DEFAULT_CUSTOM_CAP_COOLDOWN_MODE,
+    DEFAULT_CUSTOM_CAP_COOLDOWN_VALUE,
+    # Cooldown & Backoff
+    COOLDOWN_BACKOFF_TIERS,
+    COOLDOWN_BACKOFF_MAX,
+    COOLDOWN_AUTH_ERROR,
+    COOLDOWN_TRANSIENT_ERROR,
+    COOLDOWN_RATE_LIMIT_DEFAULT,
+)
+
+__all__ = [
+    # Rotation & Selection
+    "DEFAULT_ROTATION_MODE",
+    "DEFAULT_ROTATION_TOLERANCE",
+    "DEFAULT_MAX_RETRIES",
+    "DEFAULT_GLOBAL_TIMEOUT",
+    # Tier & Priority
+    "DEFAULT_TIER_PRIORITY",
+    "DEFAULT_SEQUENTIAL_FALLBACK_MULTIPLIER",
+    # Fair Cycle Rotation
+    "DEFAULT_FAIR_CYCLE_ENABLED",
+    "DEFAULT_FAIR_CYCLE_TRACKING_MODE",
+    "DEFAULT_FAIR_CYCLE_CROSS_TIER",
+    "DEFAULT_FAIR_CYCLE_DURATION",
+    "DEFAULT_EXHAUSTION_COOLDOWN_THRESHOLD",
+    # Custom Caps
+    "DEFAULT_CUSTOM_CAP_COOLDOWN_MODE",
+    "DEFAULT_CUSTOM_CAP_COOLDOWN_VALUE",
+    # Cooldown & Backoff
+    "COOLDOWN_BACKOFF_TIERS",
+    "COOLDOWN_BACKOFF_MAX",
+    "COOLDOWN_AUTH_ERROR",
+    "COOLDOWN_TRANSIENT_ERROR",
+    "COOLDOWN_RATE_LIMIT_DEFAULT",
+]