Why Pixel exists
AI coding agents learn a codebase by reading it, and most of what they read is whole files they needed a few lines of. Pixel is a local code index they query instead: the signatures of a large file, the callers of a function, the commit that changed a line, each fitted to a token budget. The benchmarks measure what that saves, and say where it does not.
Who makes it
- Navid Emad: main contributor, and publisher of this site.
- Livio Gama: created Pixel, from the idea to its structure, and remains a major contributor.
Everyone who has contributed code is listed on the repository’s contributors page.
How the project runs
- Open source: the code, its history and every review are public on GitHub, under the MIT License.
- Local by design: the index stays in
.pixel/on your machine, and the binary sends no telemetry (Privacy Policy). - Public releases: each version is a tag on
main, with its changelog and a release you can follow as a feed. - Contributions welcome: CONTRIBUTING.md lists what a pull request needs to be merged.
Contact
| For | Where |
|---|---|
| a question, an idea, or feedback on how you use Pixel | GitHub Discussions |
| a bug, with the command and its output | GitHub Issues |
| a security vulnerability, privately | a private security advisory (SECURITY.md) |
| a request about your data | as the Privacy Policy says |
There is no contact form, so the site keeps nothing you send. Every channel above is on GitHub, where you can read past threads before you write.