agent-tools AGENTS.md

Project instructions for an AI coding assistant working on agent-tools, a repository of installable coding-agent capabilities. They document the project layout, build and test commands, and where installed files come from.

In plain words
What is it for?
Use them when changing capabilities, skills, installers, build outputs, or documentation in this repository.
Why use it?
They record decisions that cannot reliably be inferred from the code, reducing mistakes such as building from the wrong directory or placing files in the wrong location.

Instructions file for CodexOpenCode

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.

agentmods
npx agentmods add instructions/kairyou/agent-tools/agents-md
Clone the repo
git clone --depth 1 https://github.com/kairyou/agent-tools

Made for: Codex, OpenCode.

Per session 564 This file is loaded in full into every session.
When invoked 564 The same file — it is already loaded in full.
Security scan A 0 findings. Scan, not verified.
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 $0.00564 $0.00564
Opus 5 $0.00282 $0.00282
Sonnet 5 $0.00113 $0.00113
Haiku 4.5 $0.00056 $0.00056

Measured 3d ago against content hash 9c2fd0e3008f, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

agent-tools 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 3d 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.

AGENTS.md · 54 lines

How it starts

The opening of the file, as written. The whole thing — 54 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Agent instructions

Conventions this repo has settled on. They live here because they cannot be inferred from the code, and getting them wrong costs a round of rework.

Build and test

Tests exercise the bundles in dist/, so run npm run build before npm test after touching anything under capabilities/.

Layout

  • Self-contained capabilities that work when copied by a Skill installer live in skills/. They may bundle scripts/, references/, and assets/ in the same Skill directory.
  • Skills centered on a specific external or business system live in skills/systems/.
  • Capabilities that require independent installation, hooks, shared runtimes, generated bundles, or Agent configuration live in capabilities/<capability>/, with per-agent adapters named <agent>-<form>.mjs.
  • Every executable installed into ~/.agent-tools by the capability installer comes from dist/; sources stay in capabilities/. Bundled scripts inside a self-contained Skill are installed with that Skill instead.
  • Capability-owned Skill templates that contain unrendered {{TOKENS}} live in capabilities/<capability>/skills/ and must not ship through npx skills add. config.default.jsonc remains the installer-merged exception from the repo root.

Settled decisions

  • No migration or legacy-compat code. Breaking a layout is fine: bump a minor version and state the one-time manual cleanup in the summary.
  • Add no new env knobs for timing: a value that follows from the caller's shape belongs in a module constant, with a comment explaining that number. The overrides already in codex-hook.mjs and claude-statusline.mjs predate this and stay unless a change removes them deliberately.
  • Never claim compatibility that has not been verified against a real gateway.
  • Usage output lines carry no branding prefix; each display surface adds its own context.

Docs

  • Never use CJK/fullwidth punctuation. Write , . : ; ( ) even in Chinese text. In - term — description lines the separator is , not -.
  • README.md, README.zh-CN.md, and both docs/ trees change together.
  • Prefer deleting a sentence over explaining it. Say what the reader must do, not how it works inside.

Read the full file on GitHub · 54 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. 3d ago First seen · 54 lines · 564 tokens per session scan A 9c2fd0e3008f

Subscribe to this mod's changes

agent-tools AGENTS.md is an instructions file published in the GitHub repository kairyou/agent-tools (172 stars, last pushed 5d ago), licensed MIT. It adds 564 tokens to every session, about $0.0028 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 2026-08-30.