Self-hosted documentation search for AI agents.

gnosis-mcp indexes your docs, git history, and crawled sites into a searchable knowledge base exposed over MCP. Zero config. SQLite by default. Hybrid FTS5 + vector with optional cross-encoder reranking.

$ pip install gnosis-mcp

Live graph → Install for your editor →

9,463

QPS at 100 docs

8.7 ms

MCP e2e mean

1.00

Hit rate @ 5

610

Tests, all green

— What it does

Your docs, indexed

Markdown, text, notebooks, TOML, CSV, JSON. Optional rST + PDF. Heading-aware chunking that never splits inside code blocks or tables.

Hybrid search

BM25 + local ONNX embeddings merged via Reciprocal Rank Fusion. Tune the fusion constant with GNOSIS_MCP_RRF_K. No API key required.

Cross-encoder rerank

Optional [reranking] extra. A 22M-param ONNX cross-encoder that re-scores the top candidates. Off by default.

Git history

Ingest commit messages as searchable context. Find the reason a line exists, not just the line itself.

Web crawl

Sitemap discovery or BFS. Robots.txt with same-host redirect guard. ETag/Last-Modified caching. Trafilatura extraction with per-page timeout.

MCP + REST on one port

9 MCP tools + 3 resources. Optional REST API on the same process with Bearer auth (timing-safe). File watcher for auto re-ingest.

— Published numbers

SQLite keyword (FTS5 + BM25), in-memory, median of 3 runs on laptop CPU. Methodology →

Corpus	QPS	p50	p95	p99
100 docs / 300 chunks	9,463	0.10 ms	0.16 ms	0.19 ms
1,000 docs / 3,000 chunks	2,768	0.29 ms	0.72 ms	0.78 ms
5,000 docs / 15,000 chunks	839	0.80 ms	2.97 ms	3.54 ms
10,000 docs / 30,000 chunks	471	1.38 ms	5.60 ms	6.29 ms

Through the full MCP stdio protocol: 8.7 ms mean, 13.0 ms p95 per tool call. RAG eval across 10 cases: Hit@5 = 1.00, MRR = 0.95, Precision@5 = 0.67. Reproduce with gnosis-mcp eval.

— vs. alternatives

	gnosis-mcp	Context7	docs-mcp-server	mcp-local-rag
Your own private docs	●	—	●	●
Self-hosted	●	hosted	●	●
Zero config install	●	●	●	●
Local embeddings, no API key	ONNX	—	opt	●
Hybrid keyword + vector (RRF)	●	—	opt	●
Cross-encoder reranker	opt	—	—	—
PostgreSQL + pgvector	●	—	—	—
Web crawl (sitemap + BFS)	●	—	●	—
Git history indexing	●	—	—	—
REST API on same port	●	—	—	—
File watcher auto re-ingest	●	—	—	—
Write tools (upsert/delete)	●	—	—	—
Published benchmarks	●	—	—	—
Built-in eval harness	●	—	—	—

Context7 indexes public library docs. gnosis-mcp indexes your own private docs. Use both.

— Get it running

Install

$ pip install gnosis-mcp

$ gnosis-mcp ingest ./docs && gnosis-mcp serve

Semantic search (local, no API key)

$ pip install gnosis-mcp[embeddings]

$ gnosis-mcp ingest ./docs --embed

Cross-encoder rerank (opt-in)

$ pip install gnosis-mcp[reranking]

$ GNOSIS_MCP_RERANK_ENABLED=true gnosis-mcp serve

Scale to PostgreSQL

$ pip install gnosis-mcp[postgres]

$ export GNOSIS_MCP_DATABASE_URL=postgresql://...

Wire it into Claude Code / Cursor / Windsurf

{
  "mcpServers": {
    "gnosis": {
      "command": "gnosis-mcp",
      "args": ["serve"]
    }
  }
}

Full editor snippets in llms-install.md.