#160 · Primary category: AI Coding Assistants

semble

agents code-search embeddings mcp mcp-server model-context-protocol retrieval

Fast and Accurate Code Search for Agents. Uses 99% fewer tokens than grep+read

Project last updated:08/26/26

GitHub Stars

6.0K

Forks

258

Contributors

9

License

MIT

Why we included this project

Coding agents waste context when they grep for keywords and read whole files just to track down a few relevant functions. Semble builds a local index instead, so a natural-language question like "how is authentication handled?" returns the matching snippets rather than a wall of source. A full codebase indexes in well under a second on CPU, with no API keys or GPU, and the project's own benchmarks claim retrieval quality close to much heavier transformer-based search at a fraction of the token cost. You can wire it in as an MCP server, a CLI hooked into AGENTS.md, or a dedicated sub-agent, and it plugs into Claude Code, Cursor, Codex, OpenCode, and similar tools. For teams that want cheaper, faster code navigation inside their existing agent setup, it is a practical drop-in.

Articles for this project

No articles for this project yet.

To suggest a topic or contribute an article, contact us.

Related projects in this category