Chat UX overhaul: sticky auto-scroll, copy/regenerate/continue actions, finish-marker handling, one-shot retry, localStorage persistence

Fix renderMdToFragment infinite loop on unmatched ** / backtick (streaming-partial freeze)

Strip debug logs; production-clean with @gradio/client@2.2.0 + break-on-complete

[DEBUG] Instrument bindComposer + refreshComposers to find post-finally button state

[DEBUG] Instrument askModel against @gradio/client@2.2.0

Upgrade @gradio/client 1.19.1 -> 2.2.0 (full rewrite, fixes iterator-not-closing hang)

Break for-await on status.stage=complete (verified via DevTools logs: 1.19.1 emits complete but never closes iterator)

[DEBUG] Instrument askModel for-await loop to diagnose _inflight hang

Revert break-on-complete (was causing issues)

Break for-await on status.complete/error — fixes inflight-stuck Send disabled after turn 1

Bump @gradio/client 1.10.0 -> 1.19.1 — fixes sse_v3 diff streaming + HF token SSE auth

Surface stream errors/timeouts even when partial content arrived (no more silent truncation)

Add Model card link to topbar

Fix unresponsive chat after several turns: in-flight gate + 90s watchdog + reset client on error

Drop NVIDIA SKU mentions, keep technical portability claim

Add lablab.ai hackathon submission attribution

Add 5-min research walkthrough video to About page

Add HF OAuth sign-in: forward visitor token to ZeroGPU backend

Tabbed UI (about=default tab, chat=second tab) + quota error handling

Restructure as single-page about-on-top + chat-below + CTA, loading status

Update askModel for new /chat schema (string in/out)

Replace C monogram with cyan-tinted Qwen logo (inlined as data URI)

Initial static frontend — calls athena129/cybersecqwen-demo via @gradio/client

initial commit