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.
npx agentmods add instructions/akshay5995/agent-skill-evals/agents-mdgit clone --depth 1 https://github.com/akshay5995/agent-skill-evalsWrote 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.
[](https://agentmods.dev/instructions/akshay5995/agent-skill-evals/agents-md)<a href="https://agentmods.dev/instructions/akshay5995/agent-skill-evals/agents-md"><img src="https://agentmods.dev/badge/instructions/akshay5995/agent-skill-evals/agents-md.svg" alt="Measured on agentmods" height="20"></a>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 | $0.00663 | $0.00663 |
| Opus 5 | $0.00331 | $0.00331 |
| Sonnet 5 | $0.00133 | $0.00133 |
| Haiku 4.5 | $0.00066 | $0.00066 |
Grade A, and why
agent-skill-evals AGENTS.md 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 4d 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.
How it starts
The opening of the file, as written. The whole thing — 72 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Agent Skill Evals Agent Guide
Why This Repo Exists
Agent Skill Evals helps people test reusable agent skills with Promptfoo. The product
should stay Promptfoo-native: users keep running promptfoo eval, and Agent Skill Evals
provides the package, providers, assertions, examples, and evidence model needed
to make those evals useful.
Repo Map
packages/promptfoo/: theagent-skill-evalspackage. Its public entry points are./agent,./assertions, and./test-generator.examples/: the public runnable workspace for real skills, real Promptfoo configs, sample projects, fixtures, and adapter evals.docs/: VitePress documentation. Keep it user-facing and grounded in current implementation.scripts/package-smoke.mjs: release-like package smoke test for the packed consumer flow.
Before Changing Code
- Use the Node version in
.nvmrcand plainpnpm. - Read the nearest implementation and docs before changing public behavior.
- Update the relevant docs in the same change whenever behavior, public APIs, examples, commands, or user-facing workflows change.
- Preserve the package boundary: do not add a root export or compatibility shim unless the user explicitly asks for a new public contract.
- Keep examples concrete and runnable. Do not add fake agent stubs to
examples/; package smoke fixtures should be generated inside the smoke harness when needed.
Product Rules
- Promptfoo is the host. Do not introduce a separate Agent Skill Evals runner for normal user flows.
- Static checks belong in core.
- Evidence is a first-class public concept. Runtime assertions should check what Agent Skill Evals can observe: files, command results, tool calls, loaded skills, usage, final output, and run details.
- Do not infer private model intent. For routing, prove that the expected skill was loaded and unrelated skills were not loaded before checking task success.
Verification
Use the smallest command that covers the change:
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.
- 4d ago First seen · 72 lines · 663 tokens per session scan A 73fc7e3cea7f
agent-skill-evals AGENTS.md is an instructions file published in the GitHub repository akshay5995/agent-skill-evals (2 stars, last pushed 1mo ago), licensed MIT. It adds 663 tokens to every session, about $0.0033 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.
Other instructions, from other repositories
caliper AGENTS.md
Instructions for edonadei/caliper, covering caliper — agent instructions, updating docs after api changes, formatting and decision docs.
skills-evals typescript.instructions.md
TypeScript conventions for this repo.
skills-evals copilot-instructions.md
Instructions for ahnafyy/skills-evals: Prefer small, focused pull requests. Run the full test suite before committing.
caliper CLAUDE.md
Instructions for edonadei/caliper, a project described as: Run your real agent with and without your skills, MCPs, and rules. See which ones actually help, and what they cost in tokens. Supports Claude Code, Codex, Pi, and Hermes.
ai-toolkit AGENTS.md
AGENTS.md instructions for pipefy/ai-toolkit, covering repository guidelines, documentation map, project structure, import namespace migration: pipefysdk → pipefy and src/pipefysdk/init.py (transitional shim).
ai-mind AGENTS.md
AGENTS.md instructions for HWYD/ai-mind, covering agents, version spec workspace continuity (mandatory), 项目定位, 事实来源优先级 and 开始大改前先读什么.