A local, offline hybrid-RAG (BM25 + vector + metadata) search server for AI agents, with a native MCP server
StrataFS exposes 4 tools with basic schemas and descriptions, but falls short of production quality. The search tool has a comprehensive parameter set (16 params) with types and descriptions, but lacks output schema documentation. The documents, health, and queue/stats tools have minimal descriptions (under 50 chars) and no visible output schemas. No error handling guidance, no parameter validation rules, and no tool annotations. Descriptions are generic and do not explain WHEN to use each tool or what the LLM should expect in responses. Parameter descriptions exist but are terse. No evidence of idempotency markers, permission gates, or recovery guidance.
Handles document retrieval requests with optional chunks and metadata
Responds to health checks with server status and version
Returns job queue statistics
Performs unified search with multiple modes and comprehensive filtering
Output schemas not documented. LLMs cannot plan downstream calls or extract required fields without knowing response structure.
Minimal descriptions for health and queue/stats tools (under 50 chars). Descriptions do not explain WHEN to call these tools or what they return.
No error handling guidance. Tools do not indicate what errors are retryable, user-fixable, or fatal, or how to recover.
search tool accepts 16 parameters but lacks constraints on numeric ranges (limit, offset, min_size, max_size) and weight values. LLMs may pass absurd values.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 43 | 2026-07-28+ | v2 |
No tool annotations (readOnlyHint, destructiveHint, idempotentHint). All tools are read-only but this is not explicitly declared.