---
title: "Agentic Evaluation Patterns - Air Market"
description: "Your agent produces something and you have no way of knowing whether it is any good. This covers ways to have outputs checked, scored, and improved before they are handed over - by critique loops, by…"
canonical: "https://airmarket.ai/app/agentic-eval"
---

[Air Market](/) / [Developer Tools](/category/devtools) / Agentic Evaluation Patterns

Skill by [GitHub](/@github)

# Agentic Evaluation Patterns

Stop shipping the first draft your agent produced

## What this does

Your agent produces something and you have no way of knowing whether it is any good. This covers ways to have outputs checked, scored, and improved before they are handed over - by critique loops, by tests, or by judging against a rubric. Reach for it when the first answer is not good enough to ship.

### Top 3 things it does for you

- Add a self-critique pass to an agent's output
- Build a pipeline that improves what it generates
- Score outputs against a rubric

## Information

**Agentic Evaluation Patterns at a glance**

|  |  |
| --- | --- |
| Kind | Skill |
| Publisher | [GitHub](/@github) |
| Category | [Developer Tools](/category/devtools) |
| Works with | Agent Skills - an open standard supported by 40+ agents |
| Size | 6 KB |
| Licence | MIT |
| Repository | [github/awesome-copilot](https://github.com/github/awesome-copilot/tree/90921cc46f4e452ea379e70884d60ec25263f16a/skills/agentic-eval) |
| GitHub stars | 38,187 (as of Nov 2026) |
| Installs | 10,274, counted by skills.sh (Nov 2026) |

## Security report

**Verified.** AIR scanned this exact version of Agentic Evaluation Patterns and flagged nothing. AIR's verdict: **safe**.

**What Agentic Evaluation Patterns's check was run against**

|  |  |
| --- | --- |
| Commit | 90921cc46f4e452ea379e70884d60ec25263f16a |
| Content hash | 2f7e4b07f3d675ee8fc2964a596ac2407264f1e5d1624e1936e35b6a82f538d5 |
| Files read | [Browse the files at this exact commit](https://github.com/github/awesome-copilot/blob/90921cc46f4e452ea379e70884d60ec25263f16a/skills/agentic-eval/SKILL.md) |

A scan is named by the exact content it read, not by a version number. A different version is a different artifact, and nothing here applies to a version AIR has not scanned.

[Read the full security report for Agentic Evaluation Patterns](/security/agentic-eval)

## About Agentic Evaluation Patterns

This collects patterns and techniques for evaluating and improving what AI agents produce, instead of accepting the first output.

It covers self-critique and reflection loops, evaluator-optimizer pipelines for generation where quality matters, test-driven code refinement workflows, and rubric-based or LLM-as-judge evaluation systems.

Use it when adding iterative improvement to agent outputs such as code, reports, or analysis, or when you want to measure and improve response quality.

## What's inside

- SKILL.md - instructions

## Questions

### Is Agentic Evaluation Patterns safe to install?

AIR scanned this exact version of Agentic Evaluation Patterns and flagged nothing. AIR's verdict: safe. The verdict is named by the content that was read, at commit 90921cc46f4e, so it cannot drift onto a version nobody scanned. The full report is at https://airmarket.ai/security/agentic-eval.

### How do I install Agentic Evaluation Patterns?

Open https://airmarket.ai/app/agentic-eval on Air Market and use the install action: Air Market hands Claude a prompt naming this exact skill, and Claude sets it up. Source, pinned to the exact commit this listing describes: https://github.com/github/awesome-copilot/tree/90921cc46f4e452ea379e70884d60ec25263f16a/skills/agentic-eval. Install from that link, not the branch.

## Related on Air Market

- [More Developer Tools listings](/category/devtools)
- [Everything by GitHub](/@github)

---

Canonical URL: <https://airmarket.ai/app/agentic-eval>