AgentSkillsCN

debug-helper

提供错误分析、日志解读与性能剖析的调试助手。当用户遇到错误、崩溃或性能问题时,可借助此工具进行排查与优化。

SKILL.md
--- frontmatter
name: debug-helper
description: Debug assistant for error analysis, log interpretation, and performance profiling. Use when user encounters errors, crashes, or performance issues.

Debug Helper

Systematic debugging workflow.

Error Analysis

  1. Read the error — Full stack trace and error message
  2. Locate — Find the source file and line
    bash
    # Search for the error origin
    grep -rn "ErrorClass\|error_function" src/
    
  3. Context — Read surrounding code (±30 lines)
  4. Reproduce — Identify minimal reproduction steps
  5. Root cause — Trace the data flow to find where it goes wrong
  6. Fix — Minimal change that addresses the root cause
  7. Verify — Run the failing case to confirm the fix

Log Analysis

  1. Read the log file or output
  2. Identify patterns: timestamps, error levels, request IDs
  3. Correlate events across log lines
  4. Summarize: what happened, when, and why

Performance Profiling

  1. Measure — Get baseline numbers first
    bash
    time <command>
    
  2. Profile — Use language-appropriate tools:
    • Node.js: --prof, clinic.js
    • Python: cProfile, py-spy
    • Go: pprof
  3. Identify — Find the hotspot (usually 1-2 functions)
  4. Optimize — Fix the bottleneck
  5. Verify — Measure again, compare with baseline