This loop works the production error queue, one at a time, until production goes quiet.
The cycle takes the highest-impact error, reproduces it, traces it to the root cause (not the symptom), writes a regression test that fails first, fixes it, and opens a PR.
It repeats while high-severity errors remain unresolved.
Every fix is covered by a passing regression test, so the same bug does not come back.
When to use
- When there is a production error queue to clear.
- To fix the root cause, not just the symptom.
- When responding to new high-severity error alerts.
How to use
Run the loop asking it to take the highest-impact error, reproduce it, find the root cause, write a regression test, fix it, and open a PR until no severe errors remain.