investigatelisted

# Investigate ## Mission Find and verify the cause before changing code. Fix the narrowest shared cause, not each visible symptom. Keep an evidence trail so conclusions are reviewable. ## Procedure ### 1. Frame the failure Record: - observed symptom and exact error text - expected behavior - reproduction steps, inputs, environment, and frequency - affected users, paths, or data - first known bad version, if available If missing context blocks useful work, ask one focused question. Otherwise continue with explicit assumptions. ### 2. Map the actual path Read repository guidance before code. Trace from entry point to symptom: 1. locate the failing symbol, route, job, command, or UI event 2. inspect callers and callees, not only the named file 3. follow data, state ownership, validation, and error handling 4. inspect tests covering the path 5. inspect recent changes to affected files Use semantic navigation when available; fall back to identifier search and targeted file reads. Do not edit during this phase. ### Historical cross-check When behavior or intent is ambiguous, inspect target-symbol history: - use blame/log to find introducing, modifying, reverting, or related commits - check accompanying tests, docs, migrations, or issue references - compare historical contract with current callers and behavior - treat commit messages as context, not proof; current evidence wins Skip broad history archaeology for obvious local errors. ### 3. Reproduce and collect evid