Debug Pattern — Problem
Pattern: Debug
Component: problem.md
Version: 1.2 | Updated: 2026-07-29
Statement
Failures get shotgun patches: first plausible cause wins, intermittent bugs “clear” after restarts, and no regression proof exists for the pre-fix behavior.
Measurable symptoms
| Symptom | How you detect it |
|---|---|
| Patch-first thrash | Multiple speculative fixes before a hypothesis list |
| Works on my machine | Fix never verified in the failing environment |
| Evidence destroyed | Restarts wipe logs needed for RCA |
| No regression net | Same defect returns within a sprint |
Root cause
Debugging under pressure rewards action over isolation. Without ranked hypotheses and a failing-env gate, teams optimize for looking busy.
What breaks when this pattern is skipped
- SEVs recur with the same fingerprint
- Hotfixes expand blast radius
- Observability exists but is never consulted before coding
Non-goals of this pattern
- Not a feature delivery track
- Not an excuse to skip tests “until later”
- Not automatic production mutation via MCP
Related
Changelog
- 1.2 (2026-07-29): Debug problem page expanded with skip-impact and operator symptoms.
Version: AIES v1.0.0✏️ Edit this page on GitHub