Haiku.rag
Agentic RAG for local and self-hosted document search: hybrid retrieval, reranking and multimodal RAG on embedded LanceDB, with Docling parsing and an MCP serve…
- Transport
- Not stated
- Package
- —
- Registry id
- —
No install snippet on purpose. A working MCP config is a command, its arguments and an environment block — the last two are where API keys live, so this catalogue never stores them and cannot publish them. Follow the link above for the authors' own instructions.
Haiku.rag indexes PDFs, web pages, and whole directories, retrieves with hybrid search, and answers with citations down to page number and section heading. Haiku.rag runs on an embedded database with open models, so documents stay on the user's machine and there is no server to operate. Haiku.rag includes visual grounding that shows the cited chunk highlighted on the original page image.
Haiku.rag provides an MCP server that exposes the database to Claude Desktop and other assistants. Haiku.rag is MIT licensed. Haiku.rag's retrieval and answer quality are tracked against public benchmarks with runnable configs.
Haiku.rag has native Pydantic AI capabilities that compose into the user's own agents.
Summary drafted from the project's own website. Every sentence is backed by text on that page and was reviewed before publishing.