High-performance MCP server for transcribing videos from 1000+ platforms (YouTube, Vimeo, TikTok, Twitter, etc.) or local video files using whisper.cpp (4-10x faster than Python whisper)
Server has 8 well-named tools with clear verb-noun patterns (transcribe_video, check_dependencies, list_transcripts, delete_transcript, etc.). Descriptions are present and substantive (avg ~150 chars), exceeding the 10-char minimum. Input schemas are visible and properly typed with enums for model selection. However, output schemas are NOT documented, LLMs cannot predict what transcribe_video returns or how to chain results. Error handling is absent: no guidance on retryability, user-fixable vs fatal errors, or recovery paths. Destructive tools (delete_transcript, delete_all_transcripts, cleanup_old_transcripts) lack confirmation patterns or dry-run support. Parameter descriptions are good but some lack format hints (e.g., 'video_id' example format not specified). No tool annotations (readOnlyHint, destructiveHint) present in schema.
Check if all required dependencies (yt-dlp, ffmpeg, whisper models) are installed
Delete transcripts older than a specified number of days. Helps manage disk space.
Delete ALL transcripts in the output directory. Use with caution - this cannot be undone!
Delete a specific transcript by video ID. This removes all associated files (txt, json, md).
Get the path and details of the most recently created/modified transcript. Useful to avoid accidentally reading old transcripts.
List all video platforms supported by yt-dlp (1000+ sites including YouTube, Vimeo, TikTok, Twitter, Facebook, Instagram, educational platforms, and more)
Output schemas not documented. LLMs cannot predict what transcribe_video, list_transcripts, or get_latest_transcript return. This breaks tool chaining and forces agents to guess field names.
Destructive tools lack confirmation or dry-run patterns. delete_all_transcripts has a 'confirm' boolean, but delete_transcript and cleanup_old_transcripts do not. Agents can accidentally destroy data without explicit user approval.
No error handling guidance. Tools do not document what errors are retryable (network timeout), user-fixable (invalid URL), or fatal (dependency missing). LLMs cannot plan recovery.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | C | 61 | 2026-07-28+ | v2 |
List all available transcripts in the output directory, sorted by modification time (newest first)
Transcribe videos from 1000+ platforms (YouTube, Vimeo, TikTok, Twitter, etc.) or local video files using whisper.cpp (4-10x faster than Python whisper!). Downloads/extracts audio and generates transcript in TXT, JSON, and Markdown formats.
Tool annotations missing. Schemas lack readOnlyHint for list_transcripts, list_supported_sites, check_dependencies; destructiveHint for delete_* tools; idempotentHint for transcribe_video. LLMs cannot infer safety properties.
Parameter format hints incomplete. 'video_id' in delete_transcript lacks example format (e.g., 'dQw4w9WgXcQ'). 'days' in cleanup_old_transcripts lacks min/max bounds. LLMs may pass invalid values.