haiku.rag
local MCP serverhaiku.rag is a local MCP server: Local-first agentic RAG with citations - hybrid search, reranking, multimodal document retrieval. Install it from pypi as haiku-rag.
Installation
Terminal
claude mcp add haiku-rag -- uvx haiku-rag
Run in your project folder. Add --scope user to use it everywhere.
Claude Desktop, Cursor & other clients
Add to claude_desktop_config.json or .cursor/mcp.json.
{
"mcpServers": {
"haiku-rag": {
"command": "uvx",
"args": [
"haiku-rag"
]
}
}
}What you need
- Runtime: Python with uv (runs with uvx).
- MCP client: Claude Code, Codex, Antigravity, Claude Desktop, Cursor or any other client that speaks MCP over stdio.
Summary
Local-first agentic RAG with citations - hybrid search, reranking, multimodal document retrieval
haiku.rag vs similar servers
| Server | Stars | Type | API key | Runtime |
|---|---|---|---|---|
| haiku.rag | 611 | Local | No | pypi |
| seo-monster | 391 | Local | Yes | pypi |
| mcp-documentation-server | 342 | Local | No | npm |
| superlocalmemory | 226 | Local | No | npm |
| sverklo | 80 | Local | No | npm |
Questions
No. The registry entry lists no keys or environment variables.
Python with uv (runs with uvx), plus an MCP client such as Claude Code, Codex, Claude Desktop or Cursor.
haiku.rag is a local MCP server distributed as the pypi package haiku-rag.
Run: claude mcp add haiku-rag -- uvx haiku-rag
Run: codex mcp add haiku-rag -- uvx haiku-rag
Similar MCP servers
NameStars
- seo-monsterio.github.avansaber/seo-monster391
- mcp-documentation-serverio.github.andrea9293/mcp-documentation-server342
- superlocalmemoryio.github.varun369/superlocalmemory226
- sverkloio.github.sverklo/sverklo80
- hypermnesia-mcpio.github.cdeust/hypermnesia-mcp73
- minirag-mcpio.github.sfrangulov/minirag-mcp67
- braveai.smithery/brave1,477
- contextio.github.neuledge/context412