スキル一覧に戻る
nguyenthienthanh

debugging

by nguyenthienthanh

Aura Frog — AI-powered structured development plugin for Claude Code Turn Claude Code into a full-fledged dev platform: Aura Frog brings 24 specialized agents, a 9-phase TDD workflow, built-in quality gates and 70+ commands so your team doesn’t need to manually draft prompts — just call the right command and follow guided instructions.

3🍴 2📅 2026年1月22日
GitHubで見るManusで実行

SKILL.md


name: debugging description: "Systematic debugging with root cause investigation. NO fixes without understanding cause first." autoInvoke: true priority: high model: sonnet triggers:

  • "bug"
  • "error"
  • "not working"
  • "broken"
  • "crash"
  • "fix" allowed-tools: Read, Grep, Glob, Bash

Debugging

Systematic debugging framework. Find root cause BEFORE fixing.


Core Principle

NO FIXES WITHOUT ROOT CAUSE INVESTIGATION FIRST

Random fixes waste time and create new bugs.


When to Use

  • Test failures
  • Bugs and errors
  • Unexpected behavior
  • Performance issues
  • Build failures
  • Before claiming "fixed"

Four-Phase Process

Phase 1: Investigate

  1. Read the error - Full message, stack trace
  2. Reproduce - Confirm the bug exists
  3. Check recent changes - git diff, git log
  4. Gather evidence - Logs, console output

Phase 2: Analyze

  1. Find working example - Similar code that works
  2. Compare - What's different?
  3. Identify pattern - Where does it diverge?

Phase 3: Hypothesize

  1. Form theory - "It breaks because..."
  2. Test minimally - Smallest change to verify
  3. Confirm - Does fix work?

Phase 4: Fix

  1. Write test first - Reproduces bug
  2. Apply fix - Single focused change
  3. Verify - Test passes, no regressions

Root Cause Tracing

When error is deep in call stack:

1. Start at error location
2. Trace backward: where did bad data come from?
3. Continue tracing until source found
4. Fix at SOURCE, not symptom

Verification Protocol

Iron Law: NO COMPLETION CLAIMS WITHOUT FRESH VERIFICATION

# Run verification
npm test  # or yarn test, pytest, etc.

# Check output
# THEN claim result

Wrong: "Should work now" Right: "Tests pass: [paste output]"


Red Flags (Stop & Follow Process)

  • "Quick fix for now"
  • "Just try changing X"
  • "It's probably X"
  • "Should work now"
  • "Seems fixed"

All mean: Return to Phase 1.


Quick Reference

Bug found
  → Phase 1: Read error, reproduce, check changes
  → Phase 2: Find working example, compare
  → Phase 3: Form hypothesis, test minimally
  → Phase 4: Write test, fix, verify

Deep stack error?
  → Trace backward to source
  → Fix at source, not symptom

About to claim fixed?
  → Run verification command
  → Read output
  → THEN claim result

References

For detailed techniques, see:

  • references/systematic-debugging.md - Full four-phase details
  • references/root-cause-tracing.md - Call stack tracing
  • references/verification.md - Verification protocols

Version: 2.0.0

スコア

総合スコア

65/100

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

+10
人気

GitHub Stars 100以上

0/15
最近の活動

1ヶ月以内に更新

+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

+5

レビュー

💬

レビュー機能は近日公開予定です