{"id":"parallel-debugging","name":"parallel-debugging","summary":"競合する仮説を用いて複雑な問題を並行調査、証拠収集、根本原因仲裁でデバッグします。複数の原因が考えられるバグのデバッグ、根本原因分析の実施、並列調査ワークフローの整理などにこのスキルを活用してください。","body":"# Parallel Debugging\n\nFramework for debugging complex issues using the Analysis of Competing Hypotheses (ACH) methodology with parallel agent investigation.\n\n## When to Use This Skill\n\n- Bug has multiple plausible root causes\n- Initial debugging attempts haven't identified the issue\n- Issue spans multiple modules or components\n- Need systematic root cause analysis with evidence\n- Want to avoid confirmation bias in debugging\n\n## Hypothesis Generation Framework\n\nGenerate hypotheses across 6 failure mode categories:\n\n### 1. Logic Error\n\n- Incorrect conditional logic (wrong operator, missing case)\n- Off-by-one errors in loops or array access\n- Missing edge case handling\n- Incorrect algorithm implementation\n\n### 2. Data Issue\n\n- Invalid or unexpected input data\n- Type mismatch or coercion error\n- Null/undefined/None where value expected\n- Encoding or serialization problem\n- Data truncation or overflow\n\n### 3. State Problem\n\n- Race condition between concurrent operations\n- Stale cache returning outdated data\n- Incorrect initialization or default values\n- Unintended mutation of shared state\n- State machine transition error\n\n### 4. Integration Failure\n\n- API contract violation (request/response mismatch)\n- Version incompatibility between components\n- Configuration mismatch between environments\n- Missing or incorrect environment variables\n- Network timeout or connection failure\n\n### 5. Resource Issue\n\n- Memory leak causing gradual degradation\n- Connection pool exhaustion\n- File descriptor or handle leak\n- Disk space or quota exceeded\n- CPU saturation from inefficient processing\n\n### 6. Environment\n\n- Missing runtime dependency\n- Wrong library or framework version\n- Platform-specific behavior difference\n- Permission or access control issue\n- Timezone or locale-related behavior\n\n## Evidence Collection Standards\n\n### What Constitutes Evidence\n\n| Evidence Type     | Strength | Example                                                         |\n| ----------------- | -------- | --------------------------------------------------------------- |\n| **Direct**        | Strong   | Code at `file.ts:42` shows `if (x > 0)` should be `if (x >= 0)` |\n| **Correlational** | Medium   | Error rate increased after commit `abc123`                      |\n| **Testimonial**   | Weak     | \"It works on my machine\"                                        |\n| **Absence**       | Variable | No null check found in the code path                            |\n\n### Citation Format\n\nAlways cite evidence with file:line references:\n\n```\n**Evidence**: The validation function at `src/validators/user.ts:87`\ndoes not check for empty strings, only null/undefined. This allows\nempty email addresses to pass validation.\n```\n\n### Confidence Levels\n\n| Level               | Criteria                                                                            |\n| ------------------- | ----------------------------------------------------------------------------------- |\n| **High (>80%)**     | Multiple direct evidence pieces, clear causal chain, no contradicting evidence      |\n| **Medium (50-80%)** | Some direct evidence, plausible causal chain, minor ambiguities                     |\n| **Low (<50%)**      | Mostly correlational evidence, incomplete causal chain, some contradicting evidence |\n\n## Result Arbitration Protocol\n\nAfter all investigators report:\n\n### Step 1: Categorize Results\n\n- **Confirmed**: High confidence, strong evidence, clear causal chain\n- **Plausible**: Medium confidence, some evidence, reasonable causal chain\n- **Falsified**: Evidence contradicts the hypothesis\n- **Inconclusive**: Insufficient evidence to confirm or falsify\n\n### Step 2: Compare Confirmed Hypotheses\n\nIf multiple hypotheses are confirmed, rank by:\n\n1. Confidence level\n2. Number of supporting evidence pieces\n3. Strength of causal chain\n4. Absence of contradicting evidence\n\n### Step 3: Determine Root Cause\n\n- If one hypothesis clearly dominates: declare as root cause\n- If multiple hypotheses are equally likely: may be compound issue (multiple contributing causes)\n- If no hypotheses confirmed: generate new hypotheses based on evidence gathered\n\n### Step 4: Validate Fix\n\nBefore declaring the bug fixed:\n\n- [ ] Fix addresses the identified root cause\n- [ ] Fix doesn't introduce new issues\n- [ ] Original reproduction case no longer fails\n- [ ] Related edge cases are covered\n- [ ] Relevant tests are added or updated","author":"@wshobson","ownerProfile":null,"authorContacts":null,"sourceUrl":"https://github.com/wshobson/agents/tree/main/plugins/agent-teams/skills/parallel-debugging","license":"MIT","category":"coding","lang":"en","tokens":860,"stars":0,"calls30d":1,"claimed":false,"visibility":"public","origin":"crawler","version":"0.1.0","createdAt":"2026-08-22","updatedAt":"2026-08-22","files":[{"path":"references/hypothesis-testing.md","size":3288,"sha256":"2042eb0d8b358bca9a098ad6d2d5e4091cfd8887e15ec6482232e87737b8c358"}],"requires":{"mcp":[],"tools":[]},"safety":{"flags":[],"scannedAt":"2026-08-22","hasScripts":false,"networkEndpoints":[]}}