From team-shinchan
Guides a 4-phase root-cause debugging process for test failures, bugs, or unexpected behavior: investigate, analyze, hypothesize, then fix one variable at a time.
How this skill is triggered — by the user, by Claude, or both
Slash command
/team-shinchan:systematic-debuggingThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
> Never guess. Investigate, analyze, hypothesize, then fix -- one variable at a time.
Never guess. Investigate, analyze, hypothesize, then fix -- one variable at a time.
git diff and git log for what changed since it last workedDo NOT attempt any fix until you can answer: "What exactly is failing and where?"
The bug is in the delta. If you cannot find the delta, your mental model of the system is wrong -- go back to Phase 1.
3+ fixes failed -- you are likely wrong about the root cause. Stop. Question every assumption you have made. Return to Phase 1 and start the investigation from scratch with fresh eyes.
| Technique | When to Use | Method |
|---|---|---|
| Root-cause tracing (5 Whys) | Symptom is clear but cause is not | Ask "why?" iteratively until you reach the origin |
| Defense-in-depth | Fix found but similar bugs could recur | Apply the fix AND add a guard (assertion, validation, type check) |
| Condition-based waiting | Timing-related failures | Replace arbitrary sleeps with explicit condition checks (poll, event, callback) |
Trying random changes hoping something works. Signs you are doing this:
If you catch yourself shotgun debugging, STOP immediately and return to Phase 1.
Guides creation, editing, and verification of skills for AI coding agents using test-driven development with subagent scenarios. Use when authoring or debugging skills.
npx claudepluginhub seokan-jeong/team-shinchan --plugin team-shinchan