メニュー
Hypothesis-driven debugging — reproduce, hypothesize, probe, locate, fix, guard. Use when something is broken and you don't yet know why.
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
| name | debug-playbook |
| description | Hypothesis-driven debugging — reproduce, hypothesize, probe, locate, fix, guard. Use when something is broken and you don't yet know why. |
The order of operations matters more than the tools.
Nothing else is useful until you can make the bug happen on demand.
List 2–3 concrete theories. For each:
If you have only one hypothesis, you're not thinking hard enough — or it's so obvious you should skip to step 4.
Add the cheapest probe that separates your top two hypotheses.
git bisect when the timeline is wide and the test is binarystrace / dtrace / perf when the surface is the runtime, not the codeResist the urge to add logs everywhere. Add one at the pivot point.
When observation singles out the hypothesis, point at the line. Not the region — the line.
Ask: why does this line break? What invariant did we assume here that isn't true?
Minimum change that resolves the root cause. Nothing adjacent.
Add a test that fails without your fix and passes with it. The test should target the condition, not mimic the scenario.
Short report:
What happened: <observable>
Root cause: <one sentence, concrete>
Why tests missed it: <honest answer>
Fix: <the change, minimum>
Regression guard: <the test>
If you can't write this report, you haven't finished debugging.
Rubric for keeping Claude Code sessions cheap without losing quality — model tiering, cache hygiene, Read discipline, MCP pruning, subagent isolation. Use when cost feels wrong, before setting a budget, or when onboarding a teammate.
Safe schema-migration patterns for systems under live traffic — expand/contract, backfill, double-write, shadow-read, online DDL. Use when authoring, reviewing, or sequencing a migration that can't take a maintenance window.
What good observability looks like — structured logs, trace spans, metrics, correlation ids, and the instrumentation rules that turn logs into answers. Use when adding a feature, reviewing a service, or debugging a prod issue.
WCAG 2.2 AA quick reference — semantics, keyboard, names, states, contrast, motion, errors. Use when building or reviewing UI components.
REST and GraphQL API design conventions — resource modeling, pagination, error envelopes, versioning, idempotency. Use when adding, changing, or reviewing endpoints.
Apply the Write/Select/Compress/Isolate framework to manage the context window. Use when the task is long, the transcript is bloated, the model is drifting, or before /holocron:handoff.