Loading ToolBench
YouTube Studio and Channel Analytics MCP Server. Provides tools for AI assistants to inspect, analyze, and diagnose YouTube channels, video performance, audience demographics, and studio analytics.
Server provides 8 well-named, read-only analytics tools with consistent verb_noun naming (get_*, list_). Descriptions are present and moderately detailed (avg ~180 chars), explaining what each tool returns. Input schemas are visible with type declarations and defaults. However, output schemas are NOT documented, LLMs cannot predict response structure. Parameter descriptions lack format constraints (e.g., date format, enum values for dimensions). Error handling is absent from tool definitions. No tool annotations (readOnlyHint, idempotentHint). Composition is sound (each tool has one responsibility), but chaining support is weak, responses don't guarantee IDs needed by downstream tools.
Fetches channel analytics broken down by a specific dimension. Returns: A ranked list of metrics for each dimension value.
Queries core YouTube Studio channel analytics for a specified timeframe. Returns: Summary metrics (total views, watch time hours, avg view duration, subscribers gained/lost, net growth, likes, shares, comments, impressions, CTR) and daily time-series data points.
Fetches high-level metadata and lifetime statistics for your YouTube channel. Returns channel title, handle, subscriber count, total lifetime views, video count, creation date, topic categories, and channel description.
Identifies and ranks your top-performing videos within a time period. Returns: Ranked list of videos with title, duration, period views, watch time, subscribers gained, and lifetime views.
Performs deep-dive analytics on a single specific video. Returns: Video metadata, daily trajectory, total period views & watch time, and traffic source breakdown for this specific video.
Output schemas not documented. LLMs cannot predict response structure, field names, or types. Tools return JSON strings without schema declarations.
Parameter constraints not formalized. 'time_period' accepts free-form strings with examples ('7d', '28d') in description instead of enum. 'dimension' parameter lists supported values in prose, not as enum constraint. LLMs may hallucinate invalid values.
No tool annotations. All tools are read-only but lack readOnlyHint annotation. Agents cannot distinguish safe tools from destructive ones without parsing descriptions.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | D | 56 | 2026-07-28+ | v2 |
Fetches comment threads and audience discussions for a video or the whole channel. Returns: List of comments with author, comment text, likes, timestamp, and sample replies.
Fetches full metadata, description, tags, and statistics for specified video IDs. Returns: Complete video details including title, description, tags, topic categories, view/like/comment counts, duration, and content settings.
Lists your channel's uploaded videos in reverse chronological order. Returns: List of videos with video ID, title, publish date, duration, views, likes, comments, and privacy status.
No error handling guidance. Tool definitions do not specify what errors can occur, whether they are retryable, or what the LLM should do next. Agents will fail silently on API errors.
Pagination support incomplete. 'list_channel_videos' accepts page_token but does not document total count, next_token, or result limits. Large result sets may exhaust context window.