skills-duplicate-detector

Duplicate detection identifies exact name matches, semantic similarities, and structural overlaps across your skill collection. The detector scans all skills, analyzes metadata, and groups related items by configurable similarity thresholds to surface perfect duplicates, near-duplicates, and functional overlaps in a consolidated report.

The detector employs exact name matching, semantic similarity analysis, and content-based structural comparison. It weights these dimensions configurably to categorize duplicates into perfect matches, near-duplicates, and functional overlaps, then outputs results with adjustable thresholds for precision control.

Execute the detector against your /skills directory to scan all SKILL.md metadata. It generates console output, Markdown reports, JSON exports, and a DUPLICATES_LOG.md file. Review the report and optionally move flagged duplicates to a safe location for consolidation.

The detector outputs console reports for immediate review, Markdown reports for documentation, optional JSON exports for programmatic processing, and a DUPLICATES_LOG.md file tracking all findings. Each format supports different workflows for reporting and backup.

Yes. The detector supports configurable similarity thresholds and weighted dimensions for name, semantic, and structural analysis. Adjust these settings to control detection sensitivity and categorization precision based on your consolidation requirements.

Review the DUPLICATES_LOG.md and reports to assess grouped skills. Back up duplicates before consolidation, then move or remove flagged items. This maintains a clean, deduplicated catalog while preserving originals until you confirm the changes.