An MCP server for managing and editing documents with support for reading, editing, and processing document content through tools, resources, and prompts.
DocumentMCP has two tools with basic but incomplete definitions. Both tools have descriptions and input schemas present, but descriptions lack LLM-optimization detail, parameter descriptions are minimal, and output schemas are undocumented. The naming is verb-forward (read_*, edit_*) which is good, but error handling is minimal and there is no guidance on recovery paths. The server also includes resources and prompts which add some capability, but the core tool definitions fall short of production-grade quality.
Edit a document by replacing a string in the documents content with a new string
Read the contents of a document and return it as a string.
Tool descriptions lack LLM-optimization detail and recovery guidance. Both descriptions are under 100 characters and do not explain WHEN to use the tool versus alternatives, or what to do on failure.
No output schema documented for either tool. LLMs cannot infer what read_doc_contents returns (string? object?) or how edit_document signals success/failure without explicit schema.
Error handling is minimal and non-actionable. When doc_id is not found, tools raise ValueError with no guidance on recovery (e.g., 'Try list_docs() to see available documents'). Agents cannot self-correct from a bare exception.
edit_document lacks a dry-run or confirmation mechanism. This is a destructive operation (modifies state) with no undo path. Agents could corrupt documents without confirmation.
Inferred effective spec: <=2025-11-25.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-22 | F | 48 | <=2025-11-25 | v2 |
| 2026-03-09 | F | 39 | - | v1 |
Parameter descriptions are generic. 'Id of the document to read' does not explain the format, valid values, or how to discover document IDs. The 'old_str' parameter does not warn that matching is exact (including whitespace), making it easy for agents to fail on invisible characters.
No idempotent guarantees documented. edit_document uses string replacement without transaction semantics, calling it twice with the same params could produce different results if the first call succeeds. Agents retry on ambiguous failures, risking unintended duplicate edits.