agent-eval-skills

agent-eval-skills is a skill for Claude Code, Codex from akshay5995/agent-skill-evals. It costs 47 tokens per session (994 once invoked), scanned A, original, MIT.

A testing toolkit for checking whether an existing coding-agent skill behaves as promised. It uses Promptfoo, a tool for running prompt-based tests, and supports behavior, routing, and role-play test types.

In plain words
What is it for?
Use it to define a skill's smallest testable promise, create a test pack, run the skill through a supported agent, and check its behavior or routing.
Why use it?
It turns vague confidence in an agent skill into repeatable tests with observable evidence, so failures can be found before the skill is relied on.

Skill for Claude CodeCodex

Written for no agent in particular: nothing here depends on one. Also seen: mentions Claude Code; mentions Codex.

Needs its repository: it reads a path above its own folder, which exists only inside the repository. The line is skill: ../skills/target-skill.

Good fit Use it to define a skill's smallest testable promise, create a test pack, run the skill through a supported agent, and check its behavior or routing.

Compare 6 skills from other repositories ↓
Install

Getting it into your agent

It runs from inside its repository, so the clone comes first — what it calls does not travel with the file alone.

Clone the repo
git clone --depth 1 https://github.com/akshay5995/agent-skill-evals
agentmods
npx agentmods add skills/akshay5995/agent-skill-evals/agent-eval-skills

Made for: Claude Code, Codex.

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.

agentmods badge for agent-eval-skills

README.md
[![agentmods](https://agentmods.dev/badge/skills/akshay5995/agent-skill-evals/agent-eval-skills/github.svg)](https://agentmods.dev/skills/akshay5995/agent-skill-evals/agent-eval-skills)
Your own site
<a href="https://agentmods.dev/skills/akshay5995/agent-skill-evals/agent-eval-skills"><img src="https://agentmods.dev/badge/skills/akshay5995/agent-skill-evals/agent-eval-skills/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.

agentmods 80×15 button for agent-eval-skills

Your own site · 80×15
<a href="https://agentmods.dev/skills/akshay5995/agent-skill-evals/agent-eval-skills"><img src="https://agentmods.dev/badge/skills/akshay5995/agent-skill-evals/agent-eval-skills.svg" alt="Reviewed on agentmods" width="80" height="20"></a>
Per session 47 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 994 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.
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.

ModelPer sessionOnce invoked
Fable 5.1 $0.00047 $0.00994
Opus 5 $0.00023 $0.00497
Sonnet 5 $0.00009 $0.00199
Haiku 4.5 $0.00005 $0.00099

Measured 10d ago against content hash 1dc188f8acd9, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-10, from the pricing page.

Security

Grade A, and why

agent-eval-skills 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 10d 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.

skills/agent-eval-skills/SKILL.md · 79 lines

How it starts

The opening of the file, as written. The whole thing — 79 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Agent Skill Evals

Build a tight red → evidence → green loop. Promptfoo is the runtime; agent-skill-evals supplies its provider, assertions, Test Pack, and static checks. Batch related reads and spend agent turns only on evidence that can change the eval.

1. Contract

Read the target SKILL.md, every file it points to, and existing project tests. State the smallest observable product promise and choose one branch: behavior, routing, or role-play.

Complete when the promise names an observable outcome and the selected branch matches it.

2. Scaffold

Detect the repository's package manager and install agent-skill-evals and promptfoo as development dependencies when absent. Run installed binaries through that package manager:

pnpm exec agent-skill-evals init --skill <path> --adapter <codex|claude-code|pi>

Adapt the command to the detected manager, such as npx agent-skill-evals for npm. Treat CLI help and generated files as the interface; source and built bundles are outside the setup path.

Confirm the selected Codex, Claude Code, or Pi CLI is installed and authenticated before the runtime eval. Package installation supplies the eval integration, not the agent executable or its credentials. For Claude Code, require CLAUDE_CODE_OAUTH_TOKEN from claude setup-token or ANTHROPIC_API_KEY; the isolated runtime cannot reuse a macOS Keychain login.

Complete when the generated config imports the package's public entry points and promptfoo eval remains the runtime command.

3. Red

Add the smallest realistic fixture. When final output cannot prove the promise, add a deterministic verifier that fails before the agent runs. Verify stable structure and facts, not incidental prose: normalize formatting, accept grammatical variation, and reject missing, misplaced, or invented facts. Exact bytes are the contract only when the target skill promises exact bytes.

When the promise is unrestricted semantic equivalence and the result is present in final output, use a native Promptfoo model-graded assertion under promptfoo.assert rather than growing a synonym list.

Read the full file on GitHub · 79 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.

  1. 10d ago First seen · 79 lines · 47 tokens per session scan A 1dc188f8acd9

Subscribe to this mod's changes

agent-eval-skills is a skill published in the GitHub repository akshay5995/agent-skill-evals (2 stars, last pushed 1mo ago), licensed MIT. It adds 47 tokens to every session and 994 once invoked, about $0.0002 per session on Opus 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 2026-08-31.

Related

Other skills, from other repositories