LEANN
Local semantic search over your own files and chats, using far less storage than typical vector databases.
- Stars
- 13.8k
- Forks
- 1.3k
- Licence
- MIT
- Language
- Python
- Last update
- 5 days ago
- Type
- Library
What it is
LEANN is a Python tool that builds a searchable index of your personal documents, such as files, emails, and chat logs, so an AI assistant can find relevant passages before answering. It is a vector database that recomputes embeddings on demand instead of storing them all, which the project says cuts storage by about 97% without accuracy loss. People use it to run private retrieval-augmented generation (RAG) on their own laptop without sending data to a cloud service.
- Index a folder of PDFs, text, and markdown files and search them by meaning
- Give a coding agent semantic search over a codebase via the MCP integration
- Search exported chat histories such as ChatGPT or Claude conversations
- Connect an LLM framework like LangChain or LlamaIndex to a local index
Great for
- Developers building a private RAG prototype on a personal machine
- People who want to semantically search their own files or notes locally
- Users of coding agents who want semantic code search through an MCP service
- Anyone with a large document collection who is short on disk space
Not ideal for
- Teams needing a hosted, multi-user database, since the README centres on a personal, single-device setup
- Users who need GPU acceleration now, as the README says it is only being considered based on survey votes
- People who want a polished end-user app with no command line, since setup uses uv and Python tooling
How to get it
- 1
Install with a package manager
You'll need: Python 3.10 or newer (3.10 through 3.14 listed in README badges) · uv package manager
Install uv
curl -LsSf https://astral.sh/uv/install.sh | shClone the repository
git clone https://github.com/StarTrail-org/LEANN.git leannCreate and activate a virtual environment
uv venv && source .venv/bin/activateInstall LEANN from PyPI
uv pip install leann - 2
Run it
- Install uv and clone the repository as shown in the README
- Create a virtual environment and activate it
- Install the leann package from PyPI
- Run the examples included in the cloned repository to see a first search
- 3
Get the most out of it
- Use the packages/leann-mcp service to connect LEANN to Claude Code for semantic code search
- Follow the Linux note and set UV_LINK_MODE=copy if leann build fails with a security violation error
- Build from source with the diskann extra if you need the DiskANN backend on macOS or Linux
- Review the benchmark and configuration docs linked in the README to tune for your data
Downloads
Strengths
- Keeps data on the local machine, with no cloud dependency described in the README
- Claims large storage savings, with a benchmark example of 60M chunks in 6GB versus 201GB
- Works with common tooling including Ollama, LangChain, and LlamaIndex, per the repository topics
Trade-offs
- Setup involves Python version requirements, uv, and optional build steps for DiskANN that can be fiddly on Linux and macOS
- Some README sections are truncated here, so the full feature set and configuration are not verified
- Accuracy and storage claims come from the project and its paper; independent verification was not part of this review
How it compares
| Project | Stars | Type | Level | Updated | Best for |
|---|---|---|---|---|---|
| LEANN (this) | 13.8k | Library | intermediate | 5 days ago | Developers building a private RAG prototype on a personal machine |
| llama_index | 52.8k | Library | intermediate | 11 days ago | Python developers building retrieval-augmented generation (RAG) apps over private documents |
| faiss | 41.2k | Library | intermediate | 5 days ago | Developers building semantic search or recommendation features over embedding vectors |
| LightRAG | 40.5k | Library | intermediate | 8 days ago | Developers building question-answering over their own document collections |
| PageIndex | 40k | Library | intermediate | 8 days ago | Developers building question-answering over long PDFs such as annual reports or manuals |
| turbovec | 17.5k | Library | intermediate | 30 days ago | Python or Rust developers building semantic search over a few million or fewer embeddings on a single machine |
Stats synced from GitHub. The assessment above was drafted with AI from the project's README on 2026-10-29 and may miss recent changes — check the README before relying on a command. MIT licence: free to use, modify and distribute, including commercially, as long as you keep the copyright notice.