Investigate Issue

Workflow

1. Define the observed behavior, expected behavior, frequency, environment, timeline, and impact. Separate reported facts from assumptions.
2. Gather the smallest useful evidence set: logs, errors, configuration, runtime state, recent changes, reproduction steps, metrics, and relevant code paths.
3. Trace execution and data flow through the affected boundary. Compare working cases and analogous repository implementations.
4. List plausible hypotheses with predicted evidence and test the most discriminating ones first.
5. Disprove alternatives where possible and identify the most likely cause, contributing factors, and evidence gaps.
6. Report findings, confidence, and the next useful action. State clearly when the evidence supports only a hypothesis.

Constraints

• Prefer direct observations and reproducible checks over intuition.
• Do not modify product behavior during an investigation-only request.
• Add temporary instrumentation only when it is safe, necessary, authorized by the task, and removable or intentionally retained.
• Do not hide contradictory evidence or overstate confidence.
• Keep investigation focused on the reported issue; avoid opportunistic refactors.

kf-fix-bug