Remedy AI code repair against the right code, at the right commit.
Prepare and assess a source-code change against a known baseline. This is the patch stage inside a wider incident-response workflow.
Exact source state prevents ambiguous repair
Remedy maps each repair to an authorised repository and exact baseline commit. The engine works in an isolated environment so the candidate change can be inspected and tested without treating a live server directory as an uncontrolled workspace.
Use the project’s own engineering gates
The repository describes the software. Its tests, build scripts, static analysis and acceptance checks provide project-specific evidence. Remedy is designed to execute the configured gates that matter to that codebase.
Reviewable delivery
A successful candidate can be placed on a dedicated repair branch and prepared for a pull or merge request where the connected workflow supports it. Approval and deployment remain governed rather than assumed.
The output of code repair
A candidate should include the starting commit, the resulting diff, the explanation and the checks that ran. It is ready for a release decision only when the agreed gates are satisfied. It is not yet evidence that a live service has recovered.
For the wider decision about recovery, configuration and source changes, see autonomous remediation. For the final production outcome, see verified repair.
Frequently asked questions
Does Remedy depend on GitHub?
No. The architecture works against mapped Git repositories and exact commits. Hosted providers are connectors, not the core repair model.
Can it repair any programming language?
Model reasoning can span many languages, but reliable repair depends on a reproducible build or test environment and meaningful deterministic gates.
From bug report to verified fix.
See how Remedy connects evidence, diagnosis, repair, deterministic checks and production verification.
Explore the complete workflow →