About this Skill
Build evaluation datasets, run and grade agent traces, and investigate failures.
Why we picked it
Covers dataset schema, metrics, judge configuration and an iterative eval-and-fix workflow.
What it helps with
- Covers dataset schema, metrics, judge configuration and an iterative eval-and-fix workflow.
Scope and use
Covers dataset schema, metrics, judge configuration and an iterative eval-and-fix workflow.
Install the suite together: this task refers to the shared workflow and other Skills. Actual cloud operations depend on your project configuration and require the appropriate authorization.
Before you use it
- The Skills work with a coding agent and the bundled references. Task execution uses agents-cli; upstream lists Python 3.11+, uv and Node.js. Local model calls require configured provider access; cloud deployment needs a Google Cloud project and may incur charges.
Sources & references
Original sources are linked below. Practical starting points are editorial suggestions from SkillHey.
