Launching today
RepoTrust
Evidence-backed scorecards for public repositories
2 followers
Evidence-backed scorecards for public repositories
2 followers
Know how a repository was built before you trust it. Evidence-backed scorecards for public repositories: AI provenance, code quality and security, scored separately. By Synaptic Cybersecurity.





Launch Team

Render Sandboxes: a computer for every agentRun untrusted code safely. Idle sandboxes cost nothing
Promoted
Developers and AppSec teams keep asking a simple question: can I trust this repository? For years we answered with stars, recent commits and a skim of the README. Those signals were weak even then, and now they're close to meaningless. AI tools can produce a polished, starred and very active repository in a weekend, and nothing on the page tells you whether anyone reviewed, tested or secured the code.
I'm not anti-AI. Plenty of AI-assisted code is excellent, and we use these tools ourselves. But how code was built is a fact you should know before you depend on it, just like its license or its test coverage. So RepoTrust gives every public repo three separate scores:
- AI provenance: how much of the code was likely AI-generated, and how sure we are.
- Code quality: complexity, duplication, tests, CI, documentation and code review.
- Security: secrets, static analysis, vulnerable dependencies and CI supply-chain risks.
We never blend them into one number, because a repo can be fully AI-written and still well engineered, or entirely hand-written and insecure. Every score links to the commits, files and lines behind it. We never run a repo's code: analysis is static, in isolated containers with no network access.
We publish our accuracy too. On a labelled set of about 700 repositories, when RepoTrust says a repo has AI-generated code it's right 99.7% of the time, and it catches about 83% of the AI-built repos.
What feedback would help most:
1. Check a repo you know well. Does the score match reality? If it doesn't, the Dispute button on every repo page is the most useful thing you can send me. Disputes improve the method for everyone.
2. Tell us what's missing for your decision. If you're about to add a dependency, what else would you want to see? License risk, maintainer activity, something else?
3. AppSec teams: would you use this in reviews or procurement? Which formats would help: an API, CI checks, badges, bulk reports?
4. Submit repos we should score. We prioritise the most-used open-source packages, but if something you depend on isn't on the board, submit it and it's usually scored within the hour.
It's built and run by Synaptic Cybersecurity, and it's free to use.