Local stdio MCP server over the Munin engine — your machine, your keys, your choice of provider. Part of the open-core local product; the hosted/managed product and the verticals are separate and closed.
Munin-mcp provides 5 tools with basic schemas and descriptions, but falls short of production quality. All tools have descriptions (10-60 chars, below the 50-200 char baseline for LLM optimization) and input schemas with type definitions. However, descriptions are generic and lack actionable context. Parameters lack detailed constraints (enums, ranges, format hints). No output schemas are documented. Error handling is absent, no recovery guidance, categorization, or actionable error messages. Tools are read-only (low risk) but lack composition clarity. The server is STDIO-only, which caps protocol readiness at 50 and limits real-world deployment.
Answer a question from the memory, optionally gathering everything known about a subject.
Gather all records for a subject entity.
Retrieve a document by its ID.
Retrieve context for a question, optionally focused on a subject.
Get the status of the Munin server.
Descriptions are too short (10-60 chars) and lack actionable context. Baseline is 50-200 chars. 'Retrieve context for a question' does not explain WHEN to use this vs munin_ask, what 'context' means, or what structure is returned.
No output schemas documented. LLMs cannot plan downstream calls or extract data without knowing what fields are returned. E.g., does munin_ask return citations, confidence scores, or raw text?
Parameters lack constraints and format hints. 'subject' and 'question' accept free-form strings with no guidance on length, format, or examples. 'pick' token is undocumented, what format? How is it obtained?
Inferred effective spec: <=2025-11-25.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | D | 50 | <=2025-11-25 | v2 |
No error handling or recovery guidance. If a question is malformed, a subject not found, or a document ID invalid, what does the tool return? How should the LLM recover?
Tool naming ambiguity: munin_retrieve_context vs munin_ask vs munin_gather_entity are semantically overlapping. LLMs will struggle to choose between them. Clarify in descriptions when to use each (e.g., 'retrieve_context for quick lookup, ask for reasoning, gather_entity for exhaustive collection').