ワンクリックで
systematic-debugging
Use when diagnosing a failure, flaky behavior, or unexpected output before attempting a fix.
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
メニュー
Use when diagnosing a failure, flaky behavior, or unexpected output before attempting a fix.
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
SOC 職業分類に基づく
Use when waiting for servers, jobs, files, tests, queues, or external processes.
Use when a bounded subtask is summarization, mechanical review, fixture inspection, proposal triage, verification, or boilerplate.
Use when locating where a symbol, function, type, command, or constant is defined or referenced, instead of grep/find/read chains.
Use when tempted to add broad abstractions, future-proofing, or extra product surface beyond the requested change.
Use before asking the user about infrastructure, preferences, conventions, prior decisions, or past incidents.
Use when evaluating reviewer feedback, review delegates, or suggested code changes.
| name | systematic-debugging |
| description | Use when diagnosing a failure, flaky behavior, or unexpected output before attempting a fix. |
| source | bundled |
Start with evidence, not guesses.
After three failed fixes, stop changing code and re-check the model of the system. Repeated patches without a new observation usually mean the cause is elsewhere.
Useful references:
references/root-cause-tracing.mdreferences/condition-based-waiting.md