debug
GitHub提供系统化的错误调试方法，涵盖复现、定位、理解、修复和验证五个步骤。针对常见错误模式给出排查建议，指导开发者高效解决代码中的意外行为和缺陷。
Trigger Scenarios
遇到程序报错或异常
需要排查代码逻辑错误
分析堆栈跟踪信息
Install
npx skills add victordibia/designing-multiagent-systems --skill debug -g -y
SKILL.md
Frontmatter
{
"name": "debug",
"triggers": "debug, error, bug, fix, not working, broken",
"description": "Systematic approach to debugging errors and unexpected behavior"
}
Debug Skill
Debugging Process
1. Reproduce
- Get the exact error message
- Find the minimal steps to trigger it
- Note the environment (Python version, OS, dependencies)
2. Locate
- Read the full stack trace (bottom = where it crashed, top = root cause)
- Find the relevant file and line
- Read surrounding code for context
3. Understand
- What was the code trying to do?
- What input caused the failure?
- What assumption was violated?
4. Fix
- Make the minimal change that fixes the issue
- Don't refactor while debugging
- Add a test that would have caught this
5. Verify
- Run the failing case again
- Run the full test suite
- Check for similar patterns elsewhere
Common Patterns
NoneType errors: Something returned None unexpectedly. Check the call chain.
Import errors: Missing dependency or circular import. Check requirements and import order.
Key/Index errors: Accessing something that doesn't exist. Add existence checks.
Type errors: Wrong type passed. Check function signatures and callers.
Workflow
grepfor the error message in codebase- Read the stack trace bottom-to-top
- Read the failing function
- Add logging/print if needed to trace values
- Fix and verify
Version History
- 1aebb31 Current 2026-07-24 16:09
