xerj
by xerj-org
XERJ is a local search engine written in Rust that indexes code, documents, logs, spreadsheets and PDFs on your own machine. It lets AI coding agents such as Claude Code and Cursor find the exact passages they need, instead of reading whole files.
- Stars
- 17k
- Licence
- Apache-2.0
- Runs on
- Your computer
- Setup
- Moderate
- Language
- Rust
- Age
- 119 days
What you can do with it
- Index a project folder and search it by keyword, meaning or both, with results pointing to file and line
- Connect it to Claude Code, Cursor or another MCP client so the agent searches instead of reading whole files
- Use it as a self-run, Elasticsearch-compatible search backend or as long-term memory for an AI agent
Could replace
- Elasticsearch
Our suggestion based on what it does, not a feature-by-feature comparison.
Make money with it
A freelance developer could offer faster, cheaper AI-assisted coding on private client codebases, or set up searchable knowledge bases for small firms.
An idea, not financial advice. Check the licence before charging for anything built on it.
Ways a business could use it
- Cut AI coding costs by letting agents fetch only the relevant code passages, so fewer tokens are used per task
- Add searchable internal documentation, logs or PDFs to a small product without paying for a hosted search service
Can you use it commercially?
Free to use, change and sell, including in paid products. Keep the copyright notice.
Not legal advice. Read the project's licence file before building a product on it.
Getting it running
Install with a one-line shell script (curl ... | sh), start the xerj server, then run xerj init and xerj autoindex on your project folder.
Follow the official instructions in the project README.
Things to check
- The --insecure quick-start switches off TLS and API-key authentication, so it should only be used by one person on a development machine
- The README asks users to send back a short field report as a community expectation, though the licence is Apache-2.0
- Token savings are the project's own figures, and it says the benefit is small or negative for popular libraries the model already knows
- Installing through a curl-to-shell script needs trust in the source
Who's talking about it
Checked 27 Oct 2026. Summary drafted by AI from the project's README.
