fix(antigravity): 🐛 add function call id fields and restrict thoughtSignature handling to gemini-3

fix(providers): 🐛 normalize JSON Schema types, clean Claude tool schemas, and fix Gemini tool conversion

fix(antigravity): 🐛 convert tool parameters to parametersJsonSchema and strip unsupported fields

fix(providers): 🐛 fix antigravity provider compatibility and async credential save

feat(antigravity): ✨ add dynamic model discovery toggle and get_valid_token helper

feat(auth): extract GoogleOAuthBase and add antigravity provider

feat(providers): ✨ add claude-sonnet-4-5 models and remove unnecessary aliasing

fix(providers): 🐛 ensure only first parallel tool call retains thoughtSignature and decouple cache/passthrough

feat(providers): ✨ add server-side thoughtSignature cache and preserve thoughtSignature handling for Gemini 3

feat(providers): ✨ support gemini 2.5/3 reasoning configs and custom budget toggle

feat(providers): add Antigravity file logging, reasoning mapping and token counting

feat(providers): ✨ add Gemini 3 thoughtSignature handling and reasoning_content separation

feat(providers): ✨ add antigravity provider and auth base

feat(auth): ✨ add sequential credential refresh queue and availability checks

Build workflow shenanigans (#9)

feat(auth): improve credential tool startup UX and lazy-load providers

fix(auth): 🐛 remove deprecated gemini models fallback

feat(auth): ✨ add headless detection, re-auth on invalid refresh tokens, and persist gemini project/tier metadata

fix(auth): 🔨 tighten oauth refresh intervals, expand refresh buffers, improve logging and add iFlow docs

build(proxy_app): 📦 include setuptools/distutils in build and clarify launcher missing .env message

feat(ui): display proxy API key and improve onboarding/security warnings

feat(ui): enhance launcher TUI and lazy-load core components

feat(ui): ✨ add interactive launcher TUI and advanced settings tool

feat(auth): add interactive onboarding and credential setup with warnings

refactor(core): resolve model id early to ensure consistent model usage

fix: Refresh qwen oauth token before returning it.

feat(core): resolve custom model ids before sending to providers

fix(providers/iflow): 🐛 stabilize iFlow streaming behavior and model handling

perf(auth): ⚡️ parallelize oauth credential initialization and deduplication

docs(readme): 📚 documentation expansion to cover new PR features(and in general)

refactor(gemini): standardize default temperature to 1

refactor(providers): standardize access token retrieval and improve setup UX

feat(gemini-cli): implement rate limit fallbacks, token counting, and stabilize onboarding

feat(setup): ensure proxy key validation runs after credential creation

feat(providers): implement detailed file logging for iFlow and Qwen Code

feat(models): support array format for simplified model definitions

refactor(providers): enhance model discovery and deduplication logic

feat(auth): enable Qwen and iFlow credential export and dynamic model discovery

refactor(auth): enhance token refresh stability and resilience

feat(gemini): implement default safety settings and image data URL support

feat(auth): enable Gemini CLI stateless authentication via environment variables

feat(auth): enable API key fallback for Qwen Code and iFlow providers

feat(auth): introduce iFlow provider integration via OAuth

build(config): 📦 update library version to 0.9

fix(core): improve error classification and prevent false key penalization

feat(core): enable configuration of maximum concurrent requests per key

refactor(core): 🔨 use static class attribute for provider skipping flag

feat(core): enable static model configuration and LiteLLM compatibility for custom providers

feat(providers): ✨ enable dynamic support for custom openai-compatible providers

refactor(client): 🔨 make credentials optional and centralize validation