Test your agent skills and fix what fails.
What it does
Creates and runs evaluations for Agent Skills, diagnoses failures, and improves skills or regression coverage when requested.
When to use it
Use when asked to test or improve a skill, add evaluation cases, capture attributed feedback, review observations, or migrate evaluations.
How to use it
Give it a skill and a request to create evaluations or improve its behavior; it writes validated evaluation files and changes the skill only when requested.
What you provide
- An existing project
What you need · 5
The evaluation workflow requires the skill-up CLI; it checks availability and provides installation instructions if it is missing.
The CLI supports macOS and Linux only; Windows is not supported.
Access · 2
This skill
skill-up
Installs the skill-up CLI
captured Skill observations
Reads captured skill observations