An automated tester for one AIWG agent, meaning an AI role configured for a particular task. It measures the agent against quality, performance, and common failure scenarios.

Use it to test a named agent, a selected test category or scenario, or the complete evaluation set with optional detailed output.

It shows whether the agent acts too early, adds irrelevant work, gets distracted, or fails to recover, as well as how fast and consistently it responds.

eval-agent — Skill for Claude Code by jmagly/aiwg

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

Any agent

npx skills add jmagly/aiwg --skill eval-agent

Clone the repo

git clone --depth 1 https://github.com/jmagly/aiwg

Made for: Claude Code.

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

README.md

[![agentmods](https://agentmods.dev/badge/skills/jmagly/aiwg/eval-agent/github.svg)](https://agentmods.dev/skills/jmagly/aiwg/eval-agent)

Your own site

<a href="https://agentmods.dev/skills/jmagly/aiwg/eval-agent"><img src="https://agentmods.dev/badge/skills/jmagly/aiwg/eval-agent/github.svg" alt="Measured on agentmods" height="20"></a>

Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.

Your own site · 80×15

<a href="https://agentmods.dev/skills/jmagly/aiwg/eval-agent"><img src="https://agentmods.dev/badge/skills/jmagly/aiwg/eval-agent.svg" alt="Reviewed on agentmods" width="80" height="20"></a>

Per session 16 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.

When invoked 992 The whole file, excluding the scripts and references it only reads on demand.

Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe. Third-party audits

NVIDIA SkillSpector pass 7 Sept 2026

How audits are shown

Origin original No closer match found in the catalogue.

Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

Model	Per session	Once invoked
Fable 5.1	$0.00016	$0.00992
Opus 5.5	$0.00006	$0.00397
Sonnet 5.5	$0.00003	$0.00198
Haiku 4.5	$0.00002	$0.00099

Measured 1mo ago against content hash 6d1492a3391f, method: parsed. Prices are Anthropic first-party input rates as of 2026-10-07, from the pricing page.

Security

Grade A, and why

eval-agent scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 1mo ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

agentic/code/addons/aiwg-evals/skills/eval-agent/SKILL.md · 147 lines

How it starts

The opening of the file, as written. The whole thing — 147 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Agent Evaluation

Run automated evaluation tests against an agent.

Research Foundation

REF-001: BP-9 - Continuous evaluation of agent performance
REF-002: KAMI benchmark methodology for failure archetype detection

Usage

/eval-agent security-architect
/eval-agent architecture-designer --category archetype
/eval-agent test-engineer --scenario grounding-test --verbose

Arguments

Argument	Required	Description
agent-name	Yes	Agent to evaluate

Options

Option	Default	Description
--category	all	Test category: archetype, performance, quality
--scenario	all	Specific scenario to run
--verbose	false	Show detailed test output
--output	stdout	Output file for results
--strict	false	Fail on any test failure

Test Categories

archetype

Tests for Roig (2025) failure archetypes:

grounding-test - Archetype 1: Premature action
substitution-test - Archetype 2: Over-helpfulness
distractor-test - Archetype 3: Context pollution
recovery-test - Archetype 4: Fragile execution

performance

latency-test - Response time benchmarks
token-test - Token efficiency
parallel-test - Concurrent execution correctness

quality

output-format - Output structure validation
tool-usage - Appropriate tool selection
scope-adherence - Stays within defined scope

Process

Load Agent: Read agent definition
Select Scenarios: Based on --category or --scenario
Setup Environment: Create test workspace
Execute Tests: Run agent against each scenario
Validate Results: Check assertions
Generate Report: Output results

Output Format

{
  "agent": "security-architect",
  "timestamp": "2025-01-15T10:30:00Z",
  "tests": {
    "grounding-test": {
      "passed": true,
      "score": 1.0,
      "details": "Read tool called before Edit",
      "duration_ms": 5000
    },
    "distractor-test": {
      "passed": false,
      "score": 0.6,
      "details": "Used staging data in output",
      "evidence": ["Found 'staging' in response"],
      "duration_ms": 3000
    }
  },
  "summary": {
    "passed": 3,
    "failed": 1,
    "total": 4,
    "score": 0.85
  }
}

Read the full file on GitHub · 147 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

1mo ago First seen · 147 lines · 16 tokens per session scan A 6d1492a3391f

Subscribe to this mod's changes

eval-agent is a skill published in the GitHub repository jmagly/aiwg (225 stars, last pushed today), licensed MIT. It adds 16 tokens to every session and 992 once invoked, about $0.0001 per session on Opus 5.5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.

Building agents Claude Code Multi-Agent Testing Workflow Automation developer-tools prompt-engineering sdlc

eval-agent