Three tools with clear verb-noun naming (get_, list_, read_) and complete input schemas. Descriptions are present and moderately detailed (100-200 chars), explaining WHAT each tool does and key parameters. However, parameter descriptions lack depth on constraints and error conditions. Output schemas are not documented, LLMs cannot predict response structure. Error handling is minimal; no guidance on recovery paths or retryability. Tool composition is sound (single responsibility), but missing idempotency hints and pagination details for list_transcripts.

• Document output schemas for all three tools. Example for get_transcript: 'Returns {filename: string, markdown: string, video_id: string, title: string, channel: string, language: string, is_generated: boolean}'. Example for list_transcripts: 'Returns {files: [{filename: string, video_id: string, title: string, created_at: ISO8601}], total: number}'.
• Add constraint descriptions to parameters. For 'languages': 'Comma-separated ISO 639-1 codes (e.g. de,en,fr). Max 5 languages. Empty string uses server default (en).' For 'filter': 'Case-insensitive substring match against filename. Max 100 chars.'
• Add error recovery guidance to get_transcript description: 'Raises ValueError if URL is invalid. Returns friendly error if video unavailable, transcripts disabled, or IP blocked. Saves successful transcripts to YT2MD_OUT directory.'
• Add pagination to list_transcripts: introduce 'limit' (default 20, max 100) and 'offset' (default 0) parameters. Return {files: [...], total: number, has_more: boolean}.
• Clarify idempotency and side effects in get_transcript: 'Idempotent, calling with same URL returns same markdown. Overwrites existing file if present. Timestamps are clickable YouTube links (e.g. https://www.youtube.com/watch?v=VIDEO_ID&t=60s).'