enforces engineering-governance checks before code changes that may be unnecessary, risky, architectural, or scope-widening. use when the user asks whether to refactor, clean up, redesign, choose a next development step, review a proposed implementation, evaluate architectural consistency, review a pull request, or prevent development drift. do not use as the primary implementation skill for routine debugging, bug fixes, feature coding, or language-specific coding unless a no-op, minimal-diff, or architecture-conflict judgment is needed.
98
100%
Does it follow best practices?
Impact
96%
1.33xAverage score across 5 eval scenarios
Passed
No known issues
{
"context": "Checks whether the agent insists on a concrete benefit for refactoring, keeps the change narrow, and avoids scope creep when a small behavior tweak is combined with cleanup pressure.",
"type": "weighted_checklist",
"checklist": [
{
"name": "Concrete reason",
"description": "Gives a concrete reason for the change from the approved set of motivations",
"max_score": 10
},
{
"name": "Smallest diff",
"description": "Prefers the smallest safe change",
"max_score": 10
},
{
"name": "Diff only if useful",
"description": "States that a diff is only justified when it improves a real benefit category",
"max_score": 10
},
{
"name": "Relevant files",
"description": "Limits the change to relevant files or areas only",
"max_score": 10
},
{
"name": "Behavior preserved",
"description": "Preserves existing behavior unless a change is required",
"max_score": 10
},
{
"name": "Focused tests",
"description": "Adds focused tests for changed behavior",
"max_score": 10
},
{
"name": "Existing patterns",
"description": "Uses existing abstractions and conventions",
"max_score": 10
},
{
"name": "Explain necessity",
"description": "Explains why each changed area is necessary",
"max_score": 10
},
{
"name": "No aesthetics",
"description": "Does not rewrite working code for aesthetics or introduce abstractions before duplication is proven",
"max_score": 10
},
{
"name": "Scope discipline",
"description": "Does not mix refactor with behavior change without saying so, and does not widen scope or rename broadly without cause",
"max_score": 10
}
]
}