MCP server for any Markdown vault as a second brain (Obsidian, Foam, Logseq, plain files)
Server has 8 well-named, read-only tools with complete input schemas and descriptions. All tools follow verb_noun naming (search_, get_, list_, find_, summarize_). Descriptions are present and adequate (100-200 chars typical). Schemas include proper types, enums, and defaults. However, output schemas are not documented, LLMs cannot predict response structure. Error handling is basic (createErrorResponse helper exists but no recovery guidance). No tool annotations (readOnlyHint, idempotentHint). Parameters lack some constraints (e.g., dateFrom/dateTo format not specified, path pattern examples only in description, not regex). Overall solid for a read-only vault tool but missing output documentation and error recovery patterns.
Scan the vault for structural gaps: wikilinks pointing to non-existent notes and notes containing unanswered questions
Retrieve the full content of a specific note by its path
Get all notes with a specific tag
Return notes not modified in N days, sorted by importance (inbound link count). Useful for spaced-repetition review.
Get the most recently modified notes
List all unique tags used across all notes
Search notes in the Obsidian vault using semantic search with optional filters
Output schemas not documented. LLMs cannot predict response structure (fields, types, nesting). Responses use createSuccessResponse(JSON.stringify(data)) but no schema is published to the client.
No tool annotations (readOnlyHint, idempotentHint). All tools are read-only and idempotent but LLMs cannot infer this from the schema. Agents may incorrectly treat them as stateful.
Date parameters (dateFrom, dateTo) lack format specification. Description says 'YYYY-MM-DD' but no JSON Schema format or pattern constraint. LLMs may pass invalid dates.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | C | 64 | 2026-07-28+ | v2 |
Get a summary of notes matching criteria
Error responses provide no recovery guidance. createErrorResponse() returns plain text errors with no actionable next steps. Pattern: 'Error: ...' does not tell LLM whether to retry, call a different tool, or ask the user.
Path parameter in get_note and search_notes accepts free-form strings with no validation. LLMs may pass invalid paths. No guidance on path format (relative vs absolute, separators, extensions).