Coastline6/transcript-agent-v2

Merge branch 'dev'

Add live interview screenshare with backend endpoints

Group interview questions by specific topic (LLM, System Design, etc.)

Bump version to 2.5.17

Merge branch 'dev'

Deduplicate interview analysis, add live streaming and Claude assessment exports

Restore HuggingFace sync workflow

Restore HuggingFace sync workflow

Disable TF oneDNN to fix SIGSEGV crash on HF free tier

Revert core files to v2.5.16 to fix container SIGSEGV (exit 139)

Remove HuggingFace sync workflow to keep source private

Remove HuggingFace sync workflow to keep source private

Disable TF oneDNN to fix SIGSEGV crash on HF free tier

Update README with all features through v2.5.17

Move network monitor to static bottom bar, widen upload column

Revert app.py UI changes, keep original progress panel layout

Fix progress panel dead space + merge fragmented diarization speakers

Merge main (v2.5.17) into dev: bring dev current with all 21 release commits (Live Transcription, report restructure, Questions tab, API interview support) while keeping dev's React UI, round-type weighting, Ollama dynamic models, OTA tag-checkout/restart fixes, and smaller UI panels

UI + Ollama + OTA fixes: dynamic Ollama model list, OTA tag-checkout & detached restart, smaller done/report panels, Gradio 6 upload-fit

Add interview round-type weighting with content-based auto-detection

Bump version to 2.5.17

Merge branch 'main' of https://github.com/jayuan101/transcript-agent

Merge simplify-analysis-options into main

Fix overly strict digit check in URL folder name parser

Merge pull request #3 from jayuan101/simplify-analysis-options

Add readable download filenames: extract candidate name from filename or URL folder

Add Questions tab/markdown export, API interview support, name binding fixes, and report dedup

Remove dev build output and Playwright driver scripts; gitignore build/

Remove unavailable claude-fable-5 default, add live model refresh

Add Live Transcription tab, Questions & Tech DOCX report, video perf fix

Add Questions Asked list and scoring-methodology notes to reports

Fix misleading "Updated!" claim on Check for Updates for source installs

Raise Interview Mode analysis output cap to avoid JSON truncation

Fix TypeError when interview-analysis JSON returns ints for text fields

Simplify Interview Mode UI, refresh AI model defaults, add Question Breakdown section

Simplify Interview Mode UI, refresh AI model defaults, add Question Breakdown section

v2.5.16: report restructure, body-language transparency, said-vs-coded coding comparison

docs: production-only README (Gradio); remove dev-only live webcam features

Support deepgram-sdk 4.1+/7.x (generated client) in _stt_deepgram

docs: rewrite README with prod vs dev editions

feat: live transcription uses Deepgram by default, with Whisper fallback

feat: add React/PrimeReact web UI with full export, language, and live-progress parity

Support deepgram-sdk 4.1+/7.x (generated client) in _stt_deepgram

Delete transcript_output directory

Merge dev: screen-name binding, Q-type breakdown, progress/ETA fixes, speaker cleanup

fix: show ETA (not just %) during the reading-names/video stage

feat: screen-name speaker binding, Q-type breakdown, progress + speaker cleanup

feat: on-screen participant name OCR + speaker mapping

remove dev docker-compose.yml — prod only via docker-compose.prod.yml

fix: v2.5.15 — auto-restart after git pull update (source install)