Haiku RAG MCP Server

Agentic RAG that answers questions about your own documents with citations to page numbers and section headings. Runs locally on an embedded database, no server required.

People who work with rag, lancedb and pydantic ai and want it reachable from Claude, Cursor, VS Code, or another MCP client. The project is written in Python.

VERIFIED ACTIVE

LAST COMMIT 2026-09-17 · ★ 609 · #4 OF 106 MAINTAINED AI & ML · VERIFIED 2026-09-18

MIT · Python servers · how we verify → /methodology

01 · Install Haiku RAG

Claude Code

claude mcp add ggozad-haiku.rag -- uvx haiku-rag

Claude Desktop / Cursor / VS Code - add to config

{
  "mcpServers": {
    "ggozad-haiku.rag": {
      "command": "uvx",
      "args": [
        "haiku-rag"
      ]
    }
  }
}

Same JSON for Cursor. For VS Code, rename the top-level key from `mcpServers` to `servers`.

Using another client? Same JSON, different key

Claude Desktop · mcpServers

Cursor · mcpServers

VS Code · servers

Windsurf · mcpServers

Zed · context_servers

Cline · mcpServers

Roo Code · mcpServers

Continue · mcpServers

LibreChat · mcpServers

Gemini CLI · mcpServers

Codex CLI · mcp_servers

Full setup guides: every client.

02 · Evidence

Security posture

What to check before giving this server access to your agent - from the registry, GitHub, and our own probes. We don't score safety; we show what's verifiable.

runs as local process (stdio) - runs on your machine with your user's permissions

license MIT - declared in the repository

pypi package haiku-rag - check the name against the project README before installing (PyPI has no namespace ownership)

registry namespace io.github.ggozad is GitHub-verified and matches the repo owner

03 · What Haiku RAG can do

Prose above is summarized from the project's README and registry record - no invented capabilities.

Latest releases

0.87.0 · 2026-09-17

Keep HTML content that precedes the first heading · Merge the analysis capability into the RAG capability

0.86.0 · 2026-09-16

Replace performed phrases with literal statements in docs · Fetch and cache remote images so HTML corpora carry picture bytes · Support email file formats · Add OpenRouter embedding, reranking and chat providers

0.85.0 · 2026-09-11

Flatten docling's inline groups on conversion · Clarify what filter scopes and what bounds a run · Recover InlineGroup text for import_document/import_documents · Update dependencies for security advisories · Document…

04 · Who maintains Haiku RAG

haiku.rag is maintained by ggozad. It's the only MCP server we track from this author; the repo dates to Oct 2024.

05 · Facts

repository: github.com/ggozad/haiku.rag
website: https://ggozad.github.io/haiku.rag/
category: AI & ML - ranked #4 of 106 actively-maintained AI & ML servers as of 2026-09-18.
release cadence: 10+ releases in the last 90 days (latest 2026-09-17)
registry: io.github.ggozad/haiku-rag (active, first published 2025-10-08 · 132 versions)
packages: pypi:haiku-rag

06 · Haiku RAG FAQ

What is Haiku RAG?

Agentic RAG that answers questions about your own documents with citations to page numbers and section headings. Runs locally on an embedded database, no server required.

Is Haiku RAG still maintained?

Yes - as of 2026-09-18, its last commit was 2026-09-17 and it shipped 10+ releases in the last 90 days. We re-verify nightly.

How do I install Haiku RAG?

Run `uvx haiku-rag`. You can also paste the ready-made client config above.

Does Haiku RAG run locally?

Yes - it's a stdio server: it runs on your machine (via uvx) with your user's permissions. Your data stays local unless the server itself calls external APIs.

07 · Alternatives to Haiku RAG

Alternatives to Haiku RAG

Maintained AI & ML servers if Haiku RAG isn't the fit.

Pairs well with

Servers that cover what Haiku RAG doesn't - only shown when the pairing reason fits the companion.

head to head · Haiku RAG vs Australian Tax MCP · Haiku RAG vs Local RAG (shinpr) · Haiku RAG vs Visual Memory MCP

More AI & ML MCP servers · Qai Consultant MCP · Omni Lpr · Unified AI System MCP Gateway · Evalview MCP · Completion Kit

More Python MCP servers · Ghost in the Droid · Securedact MCP · Decker · Maya MCP · Argentum · see all