Summary

A skill that guides agents to fix bugs by modifying the narrowest responsible layer while preserving surrounding behavior and preventing scope creep. It emphasizes reproducing failures, tracing bugs to root cause, running regression tests, and avoiding cleanup or refactoring outside the fix itself.

Detected Capabilities

Trigger Keywords

Phrases that MCP clients use to match this skill to user intent.

Use Cases

• Fix reported bugs with minimal scope and regression risk
• Make targeted behavior changes without affecting unrelated code
• Develop and verify fixes when full reproduction is expensive
• Add regression tests only for the fixed behavior
• Prevent scope creep by maintaining focus on the core issue

Quality Notes

• Core principle is well-articulated: 'narrowest responsible layer' and 'avoid cleanup, renaming, abstraction outside fix' provide clear boundaries.
• Concise checklist format (reproduce, trace, change, preserve, avoid, add, run) is easy for an agent to follow sequentially.
• Lacks concrete examples: no sample bug fix walkthrough or before/after code snippet to illustrate the methodology.
• Missing error handling guidance: what if reproduction is uneconomical? What if the narrowest layer is ambiguous?
• No explicit scope documentation: which files, directories, or test suites count as 'relevant' or 'nearest affected gate'?
• Does not describe how to verify 'preserved surrounding behavior' — no specific assertion or testing pattern provided.
• Incomplete: does not specify what 'focused proof plus nearest affected gate' means operationally (which test runner, test selection strategy, exit criteria).