---
title: "Agent Evaluator - Air Market"
description: "You have a set of agents and one of them keeps getting chosen when another should be. This goes through them and looks at how they describe themselves, which model they use, what tools they can reach…"
canonical: "https://airmarket.ai/app/eval-agents"
---

[Air Market](/) / [Developer Tools](/category/devtools) / Agent Evaluator

Skill by [Florian BRUNIAUX](/@florianbruniaux)

# Agent Evaluator

Find out why the orchestrator keeps picking the wrong agent

## What this does

You have a set of agents and one of them keeps getting chosen when another should be. This goes through them and looks at how they describe themselves, which model they use, what tools they can reach, and how their prompts are written, then points at what is causing the confusion.

### Top 3 things it does for you

- Review the agents in this project
- Find out why the wrong agent keeps getting picked
- Check whether an agent has more tools than it needs

## Information

**Agent Evaluator at a glance**

|  |  |
| --- | --- |
| Kind | Skill |
| Publisher | [Florian BRUNIAUX](/@florianbruniaux) |
| Category | [Developer Tools](/category/devtools) |
| Works with | Agent Skills - an open standard supported by 40+ agents |
| Size | 15 KB |
| Licence | CC-BY-SA-4.0 |
| Repository | [florianbruniaux/claude-code-ultimate-guide](https://github.com/florianbruniaux/claude-code-ultimate-guide/tree/73e1dec1733c6dd5b0e434890875c9cf54682fa0/examples/skills/eval-agents) |
| GitHub stars | 5,793 (as of Nov 2026) |

## Security report

**Verified.** AIR scanned this exact version of Agent Evaluator and flagged nothing. AIR's verdict: **safe**.

**What Agent Evaluator's check was run against**

|  |  |
| --- | --- |
| Commit | 73e1dec1733c6dd5b0e434890875c9cf54682fa0 |
| Content hash | ac6b0e97513d1eb47484964890307ac59c8c8bcd2b86dbbf644d80cc90d4397d |
| Files read | [Browse the files at this exact commit](https://github.com/florianbruniaux/claude-code-ultimate-guide/blob/73e1dec1733c6dd5b0e434890875c9cf54682fa0/examples/skills/eval-agents/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 Agent Evaluator](/security/eval-agents)

## About Agent Evaluator

This audits the Claude Code agents defined in .claude/agents/, looking at how specific each description is, whether the model tier fits, how tightly tools are scoped, and the quality of the system prompt.

It detects dispatch ambiguity between agents, flags over-permissive tool grants, and checks for human-in-the-loop patterns that break programmatic orchestration.

Reach for it when you are onboarding to a project that already has agents, after adding new ones to a fleet, or when an orchestrator keeps selecting the wrong agent.

## What's inside

- SKILL.md - instructions

## Questions

### Is Agent Evaluator safe to install?

AIR scanned this exact version of Agent Evaluator and flagged nothing. AIR's verdict: safe. The verdict is named by the content that was read, at commit 73e1dec1733c, so it cannot drift onto a version nobody scanned. The full report is at https://airmarket.ai/security/eval-agents.

### How do I install Agent Evaluator?

Open https://airmarket.ai/app/eval-agents 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/florianbruniaux/claude-code-ultimate-guide/tree/73e1dec1733c6dd5b0e434890875c9cf54682fa0/examples/skills/eval-agents. Install from that link, not the branch.

## Related on Air Market

- [More Developer Tools listings](/category/devtools)
- [Everything by Florian BRUNIAUX](/@florianbruniaux)

---

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