Production-ready YouTube MCP Server using Google Data API v3 - 100% reliable alternative to yt-dlp
The server has 9 tools with reasonable coverage of YouTube API functionality. Tool naming follows verb_noun conventions consistently (get_*, search_*). Descriptions are present for all tools and range from 80-250 characters, meeting the 10-1024 baseline. However, several tools lack critical schema details and parameter descriptions are incomplete. Input schemas are visible but inconsistently documented. Many parameters lack explicit type constraints (enums where applicable). Output schemas are not formally documented. Error handling is mentioned in docstrings but not reflected in actual return types. The server appears functional but falls short of production-grade tool definition quality.
Analyze engagement metrics for a YouTube video with industry benchmarks.
Extract information about a YouTube channel.
Extract information about a YouTube playlist including all videos.
Extract comments from a YouTube video with smart rate limiting.
Extract a large number of comments by making multiple smaller requests.
Extract comprehensive information about a YouTube video using YouTube Data API v3.
Output schemas not documented. Tool descriptions mention what is returned (e.g., 'List of comment threads with author, text, likes, and replies' for get_video_comments) but no formal schema definition is visible for return types. LLMs need to know the structure of responses to chain calls effectively.
search_youtube tool accepts 'search_type' as a string with description "Type of search (\"video\", \"channel\", \"playlist\")" but no enum constraint in the schema. Should be an enum type to prevent LLM from passing invalid values like 'stream' or 'user'.
Inferred effective spec: <=2025-11-25.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-22 | D | 56 | <=2025-11-25 | v2 |
| 2026-03-09 | D | 50 | - | v1 |
Extract transcript/subtitles from a YouTube video.
Search for specific text within a video's transcript.
Search YouTube for videos, channels, or playlists.
get_video_comments and get_video_comments_batch both have max_comments/batch_size parameters without explicit min/max bounds in the schema. Descriptions mention constraints ('max: 50', 'default: 20') but these are not formalized as JSON Schema minimum/maximum constraints. LLMs may not respect informal bounds.
No tool annotations visible (readOnlyHint, destructiveHint, idempotentHint). All 9 tools are READ_ONLY but this is marked in metadata, not in the tool definition itself. Modern MCP spec supports tool annotations; these should be applied to guide agent behavior.
Error handling in code shows try/except blocks that raise RuntimeError with generic messages ('Failed to extract video info', 'Failed to extract comments'). No guidance for the LLM on what to do next (retry, ask user, fetch alternatives). Should include classification (retryable, user-fixable, fatal) and recovery hints.
get_channel_info accepts 'url' parameter with verbose description of supported formats but does not document what fields are returned or how to extract a channel ID for use in downstream tools. Response structure is not documented.
search_youtube tool returns results but max_results parameter (max 50) is not enforced in schema. If an LLM passes 500, behavior is undefined. Schema should include maximum constraint.