Agent Skills
› ldayton/Dippy
› debug
debug
GitHub根据用户提供的上下文（命令、预期与实际行为、日志）定位并修复 Dippy 审批或拦截逻辑中的 Bug。遵循先写失败测试再修复的流程，确保代码质量与最小化改动。
Trigger Scenarios
用户报告命令被错误批准或阻止
提供预期与实际行为差异的调试请求
Install
npx skills add ldayton/Dippy --skill debug -g -y
SKILL.md
Frontmatter
{
"name": "debug",
"description": "Debug incorrect Dippy approval or block behavior",
"disable-model-invocation": true
}
Debug a Dippy issue based on user-provided context.
The user will provide:
- The command that was incorrectly approved or blocked
- Expected vs actual behavior
- Any relevant error messages or logs
Process
- Understand the bug from the provided context
- Search the codebase to find the relevant handler or pattern
- Write a failing test FIRST in the appropriate test file
- Run
just testand verify the test fails as expected - Fix the bug in the handler or pattern
- Run
just testuntil the test passes
Finish
When tests pass:
uv run ruff check --fix && uv run ruff format
just check MUST pass before you're done.
Restrictions
- ONLY modify files directly related to the bug
- Make minimal, targeted fixes
- Do NOT refactor or "improve" unrelated code
- Do NOT create a git commit or PR
Version History
- 2edad4a Current 2026-07-24 21:07
