This release adds 3 notable features for engineering teams evaluating rollout.
✓ No known CVEs patched in this version
Topics
+4 more
Summary
AI summaryFour new tools and three detail levels add transcript, metadata, frame extraction, and moment analysis capabilities.
Full changelog
What's New
4 New Tools
get_transcript— Quick transcript-only extraction with Whisper fallbackget_metadata— Metadata + comments + chapters (no video download needed)get_frames— Frame-only extraction (scene-change or dense 1fps sampling)analyze_moment— Deep-dive on a specific time range (burst frames + filtered transcript + OCR + mini-timeline)
Detail Levels
brief— metadata + first 10 transcript entries (fast, no download)standard— full analysis with scene-change frames (default)detailed— dense 1fps sampling, up to 60 frames, thorough OCR
Other Features
- Field filtering —
fields=["metadata","transcript"]to return only what you need - In-memory caching — 10min TTL, instant repeated queries,
forceRefreshto bypass - Dense frame sampling — 1 frame/sec for detailed analysis
- Whisper fallback transcription — @huggingface/transformers → whisper CLI → OpenAI API
- Autonomous tool selection — improved MCP instructions so Claude automatically picks the right tool
- Real example outputs —
examples/loom-demo/with 19 frames + 11 JSON files from a public Loom video
Testing
- 186 unit tests + 46 E2E tests (all passing)
- Windows ffmpeg compatibility fixes
Full changelog: https://github.com/guimatheus92/mcp-video-analyzer/compare/v0.1.0...v0.2.0
Weekly OSS security release digest.
The CVE patches and breaking changes that affected production tools this week. One email, every Sunday.
No spam, unsubscribe anytime.
Share this release
About guimatheus92/mcp-video-analyzer
MCP server for video analysis — extracts transcripts, key frames, OCR text, and annotated timelines from video URLs. Supports Loom and direct video files (.mp4, .webm). Zero auth required.
Related context
Beta — feedback welcome: [email protected]