Use when auditing a codebase for semantic duplication - functions that do the same thing but have different names or implementations. Especially useful for LLM-generated codebases where new functions are often created rather than reusing existing ones.
91
87%
Does it follow best practices?
Impact
98%
2.00xAverage score across 3 eval scenarios
Passed
No known issues
Full pipeline script usage and model selection
Extract script used
0%
100%
Catalog output flag
0%
100%
Haiku for categorization
0%
100%
Categorize not skipped
100%
100%
Prepare-category script used
0%
100%
3+ threshold applied
0%
100%
Opus for duplicate detection
0%
100%
Per-category detection
100%
100%
Generate-report script used
0%
100%
Pipeline order documented
60%
100%
Categorization phase with 3+ function threshold
Haiku model selected
0%
100%
Opus NOT used for categorization
100%
100%
categorized.json created
100%
100%
Correct categorized format
100%
100%
Purpose field present
100%
100%
All functions categorized
100%
100%
Category split performed
100%
100%
3+ threshold enforced
0%
100%
Eligible categories get files
100%
100%
Single category per function
100%
100%
Per-category duplicate detection and report generation
Opus for detection
0%
100%
Haiku NOT used for detection
100%
100%
Per-category dispatching
0%
100%
Per-category output paths
100%
100%
generate-report.sh invoked
0%
100%
Report file correct name
100%
100%
HIGH confidence prioritized
100%
100%
Verify tests before consolidating
100%
66%
Update callers step
100%
100%
Delete duplicates step
100%
100%
Run tests after consolidation
100%
100%
59389b1
Table of Contents
If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.