test-fix-loop

Automated iterative test repair runs your test suite, clusters failures by module, applies minimal implementation fixes, and re-runs until green. It requires a git-accessible repository with an identifiable test command like package.json or Makefile, and never modifies test files.

Yes, automated test repair supports CI recovery by detecting widespread test failures from recent changes, diagnosing top failing modules, and applying targeted implementation fixes. It requires permission to modify implementation code while preserving tests.

An automated test fix loop handles regressions by creating git commit checkpoints before applying fixes. If a fix causes a regression, circular fix, or non-convergence, it reverts to the previous checkpoint to protect the working tree.

The best way to diagnose and cluster test failures is using an automated loop that summarizes failures, clusters errors by module, and reads both tests and implementation code to identify root causes before applying minimal implementation changes.

No, the test-fix loop strictly preserves test files and applies changes only to implementation code. It enforces a max iteration limit and multiple termination conditions to ensure safe, controlled iterative fixes.

You should stop an automated test repair loop when tests pass, progress stalls, or a safe termination condition is reached. The loop enforces a max iteration limit and stops if it detects regressions, circular fixes, or non-convergence.