Route HuggingFace calls through backend proxy via huggingface_hub

Clean up: remove orphaned buildPrompt duplicate and huggingface_hub

Add fetchWithCorsFallback for non-streaming calls, keep apiFetch for streaming

Use huggingface_hub for HF Inference API (fixes DNS resolution from HF Spaces)

Add traceback logging to proxy endpoint, fix Content-Type check

Convert to Flask backend: server-side API proxying eliminates CORS issues

Fix HF: non-streaming chat completions with text-generation fallback, add console debug log

Fix HF API: add 30s per-proxy timeout, combineSignals helper, fix variable scoping

Show API key input for free providers too (HuggingFace token field)

Fix HuggingFace provider: use OpenAI-compatible chat completions API with streaming

Fix stop generating abort handling

Fix provider fallback and tool sources

Add tool router for search and file analysis

Prioritize free providers and auto tools

Revert "Add visual theme variants"

Add visual theme variants

Update index.html

Update index.html

Update index.html

Update index.html

Update index.html

Update index.html

initial commit