AgentSkillsCN

github-issue-resolution

通过系统化的调试与根因分析，深入调查并解决GitHub问题。

SKILL.md
--- frontmatter
name: github-issue-resolution
description: Investigates and resolves GitHub issues through systematic debugging and root cause analysis
mcp_servers:
  - github
  - git
allowed_tools:
  - github_*
  - git_*
  - Bash(git:*)
  - Read
  - Grep
  - Write
  - Edit

GitHub Issue Resolution

Expertly investigates, debugs, and resolves GitHub issues through systematic analysis and implementation.

When to Use

  • •A user reports a bug via GitHub issue
  • •Need to investigate a reported problem
  • •Implementing a fix for an open issue
  • •Creating reproduction scripts for bugs

Steps

1. Discovery

Read the GitHub issue to understand:

  • •What behavior is expected
  • •What behavior is actually happening
  • •Steps to reproduce (if provided)
  • •Environment details
  • •Any error messages or logs

2. Codebase Search

Search for relevant code:

  • •Files mentioned in the issue
  • •Error messages or keywords
  • •Related functionality
  • •Recent changes that might have introduced the bug

3. Root Cause Analysis

Analyze the issue:

  • •Trace through the code path
  • •Identify where the bug occurs
  • •Understand why it's happening
  • •Consider edge cases

4. Reproduction

Create a minimal reproduction:

  • •Write a test that fails due to the bug
  • •Or create a script that demonstrates the issue
  • •Keep it as simple as possible

5. Implementation

Implement the fix:

  • •Make the minimal necessary changes
  • •Ensure the fix addresses the root cause
  • •Consider side effects
  • •Follow existing code style

6. Verification

Verify the fix:

  • •Run the reproduction script - it should now pass
  • •Run existing tests - they should still pass
  • •Consider adding a regression test
  • •Test edge cases

7. Pull Request

Create a pull request:

  • •Reference the issue: Fixes #123
  • •Describe what was changed and why
  • •Include test results
  • •Request review if appropriate

Best Practices

  • •Always start by fully understanding the issue
  • •Don't guess - verify your assumptions with code
  • •Write tests for bugs to prevent regression
  • •Keep fixes minimal and focused
  • •Document any non-obvious changes