# semble

> semble is an open-source AI tool for coding. Local code search for coding agents that returns only the relevant snippets. It is available on CLI, macOS, Linux, Windows and self-hosted.

- **Kind:** Tool
- **Pricing:** Open source
- **Platforms:** CLI, macOS, Linux, Windows, self-hosted
- **Website:** https://minish.ai/packages/semble/introduction/
- **Repository:** https://github.com/minishlab/semble
- **GitHub stars:** 6,429
- **License:** MIT
- **Categories:** coding, developer tools
- **Listing updated:** 2026-11-14

## About

Semble indexes a repository in about half a second on average and answers queries in roughly a millisecond, all on CPU with no API keys or external services. Instead of grepping and reading whole files, an agent asks in plain language and receives just the matching chunks, which the project says uses about 99 percent fewer tokens.

Files are split into code-aware chunks with tree-sitter and scored by two retrievers: static Model2Vec embeddings and BM25. You can point it at a local path or a git URL, or several at once to search related repos together.

An interactive installer detects coding agents such as Claude Code, Cursor, Codex and Gemini CLI and sets up an MCP server, AGENTS.md instructions or a dedicated sub-agent.

## Features

- **CPU-only indexing:** No GPU, API key or external service is needed.
- **Two retrievers:** Combines Model2Vec embeddings with BM25.
- **Remote repos:** Search a local path or git URL.
- **MCP server:** Exposes search and find_related tools.
- **Interactive installer:** Detects agents and enables integrations.
- **Token-efficient output:** Returns only relevant chunks.

## Good for

- Answer where authentication is handled in a repo
- Find code similar to a given location
- Cut token use in coding-agent sessions
- Search several related repositories together

## FAQ

### Which tools does the MCP server expose?

search and find_related.

### Does it need an API key?

No; it runs on CPU.

### How accurate is it?

The project reports an NDCG@10 of 0.854 on its benchmarks.

### How do I install it?

Install uv, then run uv tool install semble and semble install.

### Is semble free?

Yes. semble is open source, so you can use and self-host it for free.

### What are the best alternatives to semble?

The closest alternatives to semble are ECC, firecrawl, ponytail, graphify and orca.

## Alternatives

- [ECC](https://ai-powered.com/tools/ecc): ECC is a freemium AI tool for ai models and llm tooling.
- [firecrawl](https://ai-powered.com/tools/firecrawl-2): firecrawl is a freemium AI tool for web search and scraping.
- [ponytail](https://ai-powered.com/tools/ponytail): ponytail is an open-source AI tool for ai models and llm tooling.
- [graphify](https://ai-powered.com/tools/graphify-2): graphify is a freemium AI tool for knowledge and memory.
- [orca](https://ai-powered.com/tools/orca): orca is an open-source AI tool for coding.
- [claude-mem](https://ai-powered.com/tools/claude-mem): claude-mem is a freemium AI tool for knowledge and memory.
- [taste-skill](https://ai-powered.com/tools/taste-skill): taste-skill is an open-source AI tool for design and media.
- [archify](https://ai-powered.com/tools/archify): archify is an open-source AI tool for design and media.

---
Source: https://ai-powered.com/tools/semble
