Remedy The repository describes the software. Its gates help prove the repair.
Language recognition is useful. Reproducible builds and meaningful deterministic checks are what make automatic program repair governable.
Examples of project verification strategies
Remedy can be designed to invoke the tools already used by an authorised project. These examples illustrate verification strategies, not guaranteed turnkey compatibility with every possible codebase.
- JavaScript / TypeScript — npm test and npm run build
- Rust — cargo test and cargo build
- Go — go test ./...
- Python — pytest
- .NET — dotnet test
- C / C++ — CMake build and ctest
- Flutter — flutter analyze and flutter test
Why reproducibility matters
An AI model may understand syntax in many languages, but a repair cannot be trusted when dependencies, build steps or expected behaviour cannot be reproduced. Remedy treats environment and gates as part of the repair contract.
Autonomous code repair with project-specific proof
Each project can define required commands, timeouts, prohibited changes and approval stages. This lets the platform extend across ecosystems without pretending every repository has the same engineering shape.
Frequently asked questions
Can Remedy fix C++ or Rust bugs?
The repair architecture can be designed around those repositories when the authorised environment and meaningful build and test gates are reproducible. That is not a guarantee for every project.
Does Remedy invent its own tests?
It can propose targeted regression checks, but the project’s approved deterministic gates remain central to judging a repair.
From bug report to verified fix.
See how Remedy connects evidence, diagnosis, repair, deterministic checks and production verification.
Explore the complete workflow →