Self-healing Loop
Self-healing Loop
Section titled “Self-healing Loop”The standard is not “code changed.” The standard is runtime evidence.
- Observe the symptom.
- Collect evidence.
- State the hypothesis.
- Make the smallest fix.
- Restart, refresh, or deploy through SuperDev when needed.
- Collect evidence again.
- Report the result and remaining risk.
Evidence By Scenario
Section titled “Evidence By Scenario”- Backend: service status, log context, readiness, error windows, breakpoint variables.
- Frontend: screenshot, stable selectors, console logs, network requests, interaction result.
- Deployment: pipeline run status, step logs, artifacts, rollback readiness.
An agent that cannot show evidence should say what evidence is missing and collect it next.