# codesearch MCP server

> Lets AI agents run hybrid semantic and lexical code search across multiple indexed repositories, with symbol navigation and AST-aware chunks.

- Repository: https://github.com/flupkede/codesearch
- Page: https://hostedmcp.org/servers/flupkede/codesearch
- Transport: —
- Stars: 104
- License: Apache-2.0

## Tools (20)

- `query`
- `mode`
- `filter_path`
- `file_glob`
- `language`
- `regex`
- `phrase`
- `compact`
- `limit`
- `project`
- `group`
- `symbol`
- `kind`
- `definition_kind`
- `target`
- `chunk_id`
- `context_lines`
- `symbol_name`
- `file`
- `line`

## FAQ

### Does codesearch need a GPU or Docker?

No. It runs on CPU only with a footprint of a few hundred MB and no Docker requirement.

### Which MCP clients work with codesearch?

It works with OpenCode, Claude Code, Cursor and any MCP client, via stdio or HTTP.

### Can I change the embedding model after indexing?

Yes, but it requires a full reindex because embedding dimensions and vector spaces are model-specific.
