A markdown document indexing and search server with full-text search, vector semantic search, and hybrid search capabilities. Provides MCP tools for querying a local knowledge base of markdown files with optional LLM-powered embeddings.
Three search-focused tools with complete input schemas and reasonable descriptions. All tools have verb-noun naming (search, vsearch, query) and documented parameters with types and defaults. However, descriptions lack depth on when to use each tool vs. alternatives, output schemas are not formally documented, and error handling guidance is absent. Parameter descriptions are adequate but generic. No tool annotations (readOnlyHint) despite all being read-only operations.
Hybrid search using both keywords and semantic meaning (RRF). Best for most queries. If 'full_file_returned' is true, 'snippet' contains the complete document content and 'get_document' is not required.
Full text search using BM25. Returns a JSON list of matches. Use context_lines to see surrounding text. If 'full_file_returned' is true, 'snippet' contains the complete document content and 'get_document' is not required.
Semantic search using vector embeddings. Returns the matched text chunk. If 'full_file_returned' is true, 'snippet' contains the complete document content and 'get_document' is not required.
Output schemas not documented. Tool descriptions mention 'Returns a JSON list' but do not specify field names, types, or structure. LLMs cannot plan downstream operations without knowing what fields to extract.
No tool differentiation guidance. Descriptions do not explain when to use 'search' (BM25) vs 'vsearch' (semantic) vs 'query' (hybrid RRF). LLMs will guess or try all three, wasting tokens and latency.
No error handling guidance. Descriptions do not explain what happens on invalid queries, missing files, or empty results. LLMs have no recovery path if a search fails.
Tool annotations missing. All three tools are read-only but lack readOnlyHint annotation. Agents cannot distinguish safe tools from destructive ones without explicit hints.
Inferred effective spec: <=2025-11-25.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | D | 58 | <=2025-11-25 | v2 |
Parameter descriptions lack format/constraint details. 'context_lines' and 'limit' accept numbers but do not specify min/max ranges. LLMs may pass absurd values (limit=999999, context_lines=-10).