An MCP server that gives your coding agent local search. Point your agent at lilbee and it reads your real code and docs before it answers, with file and line citations.
lilbee speaks the Model Context Protocol, so any MCP-aware agent can search the files and code on your own machine. Retrieval stays local. Nothing is sent to a cloud service.
Private by default. Your code stays on your machine. No telemetry, no account. A cloud model runs only when you pick one.
why retrieval before the answer
An agent that guesses from memory is confident and wrong. With lilbee in the loop, the agent first retrieves the relevant passages from your indexed code and docs. It answers from what it found and cites the file and line, so you can check it. When there is no good match, it says so instead of inventing one.
what the server gives the agent
search
Ranked retrieval over a project library, through the same pipeline the terminal app uses. Results come back in a shape an agent can act on.
add & sync
The agent can pull a file or folder into a library and keep the index current, so what it searches matches what is on disk.
crawl
Fetch a web page or a docs site into the library as markdown, then search it offline alongside your local files.
wiki & datasets
Build a wiki over a library, and export or import a library as a dataset to move it between machines.
it runs the models too
lilbee is not only the retrieval layer. It is a complete model manager: it downloads and runs the models itself on Metal, Vulkan, or CUDA. One program covers both the search and the model behind it. Already running Ollama or LM Studio? Point lilbee at your setup and keep it.
questions
What is the lilbee MCP server?
lilbee runs as a Model Context Protocol server, so any MCP-aware coding agent can search your local files and code through it. The agent reads the real source before it answers and cites the file and line.
Which agents work with it?
Any MCP-aware coding agent. lilbee exposes standard MCP tools, so you add it the same way you add any other MCP server in your agent's configuration.
What tools does it expose?
Search, add files, sync the index, crawl a web page, manage per-project libraries, build a wiki over a library, and export or import a dataset.
Does my code leave my machine?
No. Indexing and retrieval run locally. lilbee uses a cloud model only if you pick one; the search itself stays on your machine.
Can lilbee run the models too?
Yes. It is a complete model manager and runs models on your GPU, or it works with your existing Ollama or LM Studio setup.