スキル一覧に戻る
BaxterCooper

systematic-debugging

by BaxterCooper

0🍴 0📅 2026年1月20日
GitHubで見るManusで実行

SKILL.md


name: systematic-debugging description: Invoke when debugging failures. MANDATORY when 3+ fix attempts have failed. Enforces root cause investigation before fixes. version: 1.0.0

Systematic Debugging Skill

Core Principle: No fixes without root cause investigation first.

The 3+ Fix Rule

[!CRITICAL] If you have attempted 3 or more fixes without success, STOP. This indicates an architectural problem, not an isolated bug. Escalate to user with analysis.


Four Phases

Phase 1: Investigate

Before attempting ANY fix:

ActionPurpose
Read error messages completelyErrors tell you what's wrong
Reproduce the issueConfirm you're fixing the right thing
Check recent changesWhat changed before it broke?
Trace data flowFollow the data from input to error
Gather diagnostic evidenceLogs, stack traces, state dumps

Output: Clear statement of what's happening and where

Phase 2: Analyze

Compare broken state to working state:

ActionPurpose
Find working examplesWhat does correct behavior look like?
Compare differencesWhat's different between working and broken?
Check assumptionsWhat are you assuming that might be wrong?
Review dependenciesWhat does this code depend on?

Output: List of differences between working and broken states

Phase 3: Hypothesize

Form a specific, testable hypothesis:

Good HypothesisBad Hypothesis
"The null check on line 42 doesn't handle undefined""Something's wrong with the data"
"The API returns 404 when ID contains spaces""The API is broken"
"State updates before render completes""It's a timing issue"

Test minimally: Change ONE thing to test the hypothesis

Output: Specific hypothesis with test plan

Phase 4: Implement

Only after phases 1-3:

  1. Write failing test that demonstrates the bug
  2. Implement single fix addressing root cause
  3. Verify fix works
  4. Verify no regressions (run full test suite)

Output: Fix with evidence it works


Red Flags: Restart the Process

If you catch yourself doing any of these, STOP and restart from Phase 1:

Red FlagWhat's Wrong
Guessing at fixesSkipped investigation
"Let me try this..."No hypothesis
Multiple changes at onceCan't isolate cause
"One more fix should do it"Sunk cost fallacy
Pretending to understandIncomplete analysis

The 3+ Fix Escalation

After 3 failed fix attempts:

┌─────────────────────────────────────────────┐
│ 3+ Fix Attempts Failed                       │
├─────────────────────────────────────────────┤
│ This pattern indicates:                      │
│ - Fundamental design problem                 │
│ - Missing understanding of the system        │
│ - Wrong level of abstraction                 │
│                                              │
│ Recommended actions:                         │
│ 1. Document what was tried                   │
│ 2. Describe current understanding            │
│ 3. Identify knowledge gaps                   │
│ 4. Escalate to user/architect                │
└─────────────────────────────────────────────┘

Debugging Report Template

After investigation, report:

issue:
  symptom: "[What the user sees]"
  location: "[File:line or component]"
  reproducibility: "[Always/Sometimes/Rarely]"

investigation:
  error_message: "[Exact error]"
  stack_trace: "[Relevant frames]"
  recent_changes: "[What changed]"
  data_state: "[Relevant data at failure point]"

analysis:
  working_comparison: "[What works vs what doesn't]"
  differences_found:
    - "[Difference 1]"
    - "[Difference 2]"
  assumptions_checked:
    - "[Assumption]: [Valid/Invalid]"

hypothesis:
  statement: "[Specific hypothesis]"
  test_plan: "[How to test]"
  test_result: "[What happened]"

fix:
  change: "[What was changed]"
  verification: "[How verified]"
  regression_check: "[Test results]"

Integration with Orchestration

This skill is invoked when:

  • orchestration detects 3+ failed fix attempts
  • User explicitly requests debugging help
  • verification skill repeatedly fails

Chain: orchestration → (3+ failures) → systematic-debuggingexecutor


Deviation Logging

Skipped debugging logged:

- id: [auto]
  timestamp: [now]
  expected: "Systematic debugging after 3+ failures"
  actual: "Continued attempting fixes without investigation"
  root_cause: skipped_debugging
  fix: |
    Enforce systematic-debugging skill after 3 failures.
    Add counter to orchestration skill.

スコア

総合スコア

40/100

リポジトリの品質指標に基づく評価

SKILL.md

SKILL.mdファイルが含まれている

+20
LICENSE

ライセンスが設定されている

0/10
説明文

100文字以上の説明がある

0/10
人気

GitHub Stars 100以上

0/15
最近の活動

3ヶ月以内に更新がある

0/10
フォーク

10回以上フォークされている

0/5
Issue管理

オープンIssueが50未満

+5
言語

プログラミング言語が設定されている

+5
タグ

1つ以上のタグが設定されている

0/5

レビュー

💬

レビュー機能は近日公開予定です