Inspiration

AI coding agents can make tests pass without genuinely fixing the underlying bug. We built TrueFix to address the gap between a green pipeline and a trustworthy fix.

What it does

TrueFix is a deterministic fix-integrity verification system that detects weakened tests, validates code changes, and checks evidence integrity through tamper detection.

How we built it

We used GitLab repositories, merge requests, CI/CD pipelines, automated verification jobs, a deployed scorecard, and receipt/hash validation to make results inspectable and reproducible.

Challenges

A key challenge was distinguishing a genuinely validated change from a misleading green test result. We designed negative and positive validation runs to demonstrate the difference.

What we learned

Reliable AI-assisted development needs independent verification and auditable evidence, not just successful test execution.

What's next

Validate the custom Repairer, Skeptic, and Governor flows end-to-end on the platform.

Built With

  • automation
  • ci/cd
  • devops
  • gitlab
  • python
  • softwaretesting
  • yaml
Share this project:

Updates

Submission history