Loop — Iterative Test-Fix Cycle#

Purpose#

Iteratively improve code by running tests, fixing failures, and repeating until all tests pass. Enforces a maximum iteration limit to prevent infinite loops.

Workflow#

1. Initialize: Set iteration counter to 0. Maximum iterations: 10.
2. Run Tests: Execute the project's test suite:
   • Detect test runner from project config (package.json, Makefile, pyproject.toml).
   • Run the full suite or targeted tests if specified by the user.
3. Evaluate Results:
   • If all tests pass: report success and exit.
   • If tests fail: proceed to fix step.
   • If iteration count >= 10: stop and report remaining failures.
4. Fix Failures:
   • Analyze each failing test's error message and stack trace.
   • Identify the root cause in the source code (not the test).
   • Apply the minimal fix needed.
   • Increment iteration counter.
5. Repeat: Go back to step 2.

Progress Reporting#

After each iteration, report:

• Iteration number (e.g., "Iteration 3/10")
• Tests passed / total
• What was fixed in this iteration
• Remaining failures

Guidelines#

• Fix source code, not tests (unless the test itself is clearly wrong).
• Apply one fix per iteration to keep changes traceable.
• If the same test fails 3 times with different fixes, flag it for manual review.
• Never disable or skip tests to make them "pass."
• After completion, show a summary of all changes made across iterations.
• If no test runner is found, ask the user how to run tests before starting.