@autodev/codebase CLI
CLI and MCP server for vector-based semantic code search: local Ollama embeddings, call graphs and file outlines
- 121 ★ on the repository
- Last commit
Extended access
Why this level
- Indexes the whole repository and, with a cloud embedding provider, sends code fragments outside.
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add cli/autodev-codebaseDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewExtended accessThe tool runs code or makes network calls and only reads external data.
- Indexes the whole repository and, with a cloud embedding provider, sends code fragments outside.
This is third-party code. Review the repository files before installing.
What it does
@autodev/codebase indexes a repository into the Qdrant vector store and searches by meaning rather than keywords. The query `codebase search "user manage"` returns matching code blocks with line numbers. Parsing uses Tree-sitter for more than 40 languages, and the index updates as files change. Beyond search, commands build a call graph (`codebase call --query="app,addUser"` shows connections and chains between functions) and file outlines with short summaries through `codebase outline ... --summarize`. For agents an MCP server starts with `codebase index --serve --port=3001` at `http://localhost:3001/mcp`. For IDEs that support only stdio there is an adapter `codebase stdio --server-url=...`. Embedding providers are Ollama (fully local), OpenAI, Jina, Gemini, Mistral, OpenRouter and Vercel.
Who it is for. Developers of large repositories who need meaning-based search for an agent and IDE, ideally without sending code to external services.
Good fit when
- Code is found by the meaning of a query like «user manage», with no exact word needed.
- An agent needs semantic project search through MCP.
- A call graph between functions is needed to trace an execution path.
- Embeddings must be computed locally through Ollama so code never leaves the machine.
Not a fit when
- A small project where name search through `rg` is enough.
- Docker with Qdrant cannot run on the machine.
- You need a ready cloud solution with no setup.
Example request
Find the code that manages users in this project and show the call graph around it.Limitations
Ollama with the `nomic-embed-text` model (or another embedding provider), `ripgrep` and Qdrant running in Docker are required. Setup steps: `brew install ollama ripgrep`, `ollama serve`, `ollama pull nomic-embed-text`, then a Qdrant container and `npm install -g @autodev/codebase`. The last commit in the repository dates from July 2026. The license is MIT, though it is missing from repository metadata. With a cloud embedding provider code fragments go to its server, and the MCP server listens on a local port without authorization.
How to disable. Stop `codebase index --serve`, uninstall the package with `npm uninstall -g @autodev/codebase`, remove the Qdrant container and the `codebase` entry in the IDE MCP settings.
FAQ
How do I install it?
Run `npm install -g @autodev/codebase`, then `codebase config --set embedderProvider=ollama,embedderModelId=nomic-embed-text`. Start Ollama and Qdrant first.
How do I start the MCP server?
With `codebase index --serve --port=3001`. Add the address `http://localhost:3001/mcp` to the client configuration.
What if the IDE accepts only stdio?
Start the server, then the adapter `codebase stdio --server-url=http://localhost:3001/mcp` and register it as the command.
Can it be tried without my own project?
Yes. Use the `--demo` mode: `codebase index --demo`, then `codebase search "user greet" --demo`.
Does it work offline?
Yes. With Ollama embeddings and a local Qdrant, indexing and search run entirely on the machine.
Discussion
No questions or reviews yet.
Related
CLI toolsA local semantic code index: the agent searches by meaning, then follows hits into definitions, callers and dependency paths
A local codebase index for agents: hybrid search, dependency graphs and change impact analysis
The CodeAlive MCP server: the agent searches large codebases by meaning and exactly and follows call relations without a full read
codesearch searches code across several repositories with hybrid retrieval and runs fully offline, without a GPU or Docker.
A local MCP server and CLI for agents: semantic code search, call graphs and change-impact analysis
A Rust MCP server and CLI with a call graph and hybrid semantic code search for Claude Code and Codex
README in short
The project combines an indexer, search, a call graph, file outlines and an MCP server in one console utility with no GUI.