Find tools › Skills › Make your AI assistant smarter

grade-iterate

Skill · by claude-skills

Grade a launched managed agent against quality benchmarks and iterate to improve its output.

Some setupInstalled 4 times

What it costs

Free and open source.

Use it to

  • Grading an agent against rubrics
  • Testing agents with evaluation scaffolds
  • Evaluating agent response quality

How to start

  1. Open the skill's folder on GitHub and skim SKILL.md, so you know what it will do.

  2. Add the folder to your assistant. Claude Code: copy it into ~/.claude/skills/. Codex: into .agents/skills/. Claude app: zip the folder and upload it under Settings, Capabilities, Skills.

  3. Ask for the task in plain words. The assistant loads the skill by itself when your request matches. Try: "Grade my newly launched agent against the rubric to see if it is good enough."

Before you install

  • Includes scriptsIt can run programs on your computer. Skim the scripts folder first.
  • From a collection we chose by handThe collection is well known; this skill's own text was indexed automatically.
  • License: MITYou are allowed to use it under these terms.
  • Last updated 2026-11-17

Similar tools

Descriptions are written in plain English from each project's own documentation and refreshed by our agent. Spotted a mistake? · Open in Find tools · © 2026 Which AI Map