MCP server that wraps the Zenn CLI (npx zenn) for use by MCP clients
zenn-mcp provides 5 tools with consistent Zod-based schemas and descriptions. All tools have descriptions (10 - 100 chars) and input parameters with type definitions. However, descriptions are minimal (avg 80 chars vs. baseline 194), lack WHEN/WHY context, and omit error recovery guidance. Output schemas are not documented, LLMs cannot predict response structure. No tool annotations (readOnlyHint/destructiveHint). Parameter descriptions are sparse; e.g., 'Article slug' lacks format constraints (12-50 chars mentioned in tool description, not param description). No pagination support for list tools. Error handling exists but returns generic messages without recovery hints.
Initialize a new Zenn project in the given directory (runs `npx zenn init`).
List Zenn articles in the given project directory (runs `npx zenn list:articles`).
List Zenn books in the given project directory (runs `npx zenn list:books`).
Create a new Zenn article in the given project directory (runs `npx zenn new:article`).
Create a new Zenn book in the given project directory (runs `npx zenn new:book`).
Output schemas not documented. LLMs cannot predict response structure (fields, types, pagination). Baseline: 100% of A+ tools document return types.
Descriptions lack WHEN/WHY context and error recovery guidance. Avg 80 chars vs. baseline 194. E.g., 'Initialize a new Zenn project' does not explain when to call it or what happens on failure.
Parameter descriptions are minimal. 'Article slug' lacks format constraints; constraint '12-50 chars' is in tool description, not param description. LLMs cannot read tool descriptions for param validation.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | D | 55 | 2026-07-28+ | v2 |
List tools (zenn_list_articles, zenn_list_books) lack pagination (limit, offset, cursor). Large result sets will blow context window. Baseline: paginated-result pattern required for list tools.
No tool annotations (readOnlyHint, destructiveHint, idempotentHint). LLMs cannot distinguish safe reads from destructive writes without explicit hints.