This is a structured workflow for fixing bugs without letting scope creep turn a small patch into an architecture overhaul. It gates you at the start: concrete issue, localized fix path, straightforward verification. If any of those fail, or if root cause stays murky after investigation, it tells you to stop and escalate to brainstorm or write-plan. The workflow itself is methodical: reproduce, diagnose root cause, classify as small or risky, implement minimal change, verify with increasing scope, then document what broke and why your fix works. It's opinionated about keeping blast radius tight and not guessing when context is missing. Good for runtime errors and test regressions where you have real signal to work from.
- skills.sh installs
- 207
- GitHub stars
- 48
- last updated
- Jun 3, 2026
Install
$ npx -y skills add buiducnhat/agent-skills --skill fix --agent claude-codeInstalls into .claude/skills of the current project.
Files
Fix
Scope Gate (Required Before Starting)
Use this skill only when all conditions below are true:
-
Concrete issue signal exists
- Clear failing behavior, error, or regression is identified
- Enough evidence exists to begin diagnosis (logs, stack trace, failing test, or reproducible steps)
-
Likely localized fix path
- Issue appears constrained to a small area
- No expected cross-system redesign or migration
-
Low-to-moderate architectural risk
- Fix can be made without foundational changes
- No phased rollout/feature-flag strategy is expected
-
Straightforward verification path
- Reproduction and post-fix checks can be run in practical scope
- Success criteria can be validated directly
If any condition fails, escalate to brainstorm or write-plan.
Hard Stop Escalation Criteria
Immediately stop and switch to planning if any of these occur:
Sponsors
Details
- Repository
- github.com/buiducnhat/agent-skills
- Categories
- AI & Agent Building, Debugging
- First seen
- Jun 3, 2026
