NVIDIA/SkillEvaluator
Multi-tier evaluation framework for AI agent skills and plugins
- Daily
- #5
- Today
- +312
- Stars
- 3.9k
- Daily peak
- #5
- Days on chart
- 2
What it is
SkillEvaluator is a multi-tier framework that checks AI agent skills and plugins for quality, security, overlap, and compatibility, and measures how they affect task results. It supports keyless static quality checks as well as sandboxed live agent evaluations.
Who it is for
Developers building or validating skills and plugins for coding agents such as Claude Code and Codex.
Why it is trending
It gained 969 stars today to reach 3,179 total stars, following the SkillEvaluator 0.9.0 release on November 13, 2026. It stands out by combining static quality gates and deduplication with live agent runs that compare behavior with and without a skill.
Main features
- Keyless static quality and harness compatibility checks
- Semantic overlap and deduplication across skill catalogs
- Live agent evaluation in Docker sandboxes
- Automated CI workflow setup for GitHub repositories
Summarized on November 17, 2026 from the README and repository data. It can be wrong; check the source before relying on it. Report a wrong summary