agent-levers AGENTS.md

Repository instructions for agent-levers, a tool that gives coding agents a repeatable plan, work, check, and review cycle using shared files.

In plain words
What is it for?
Use them to create, run, and track structured work items called levers, with their details and validation stored in the project.
Why use it?
They give the agent consistent project rules and keep work organized across different coding-agent tools.

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/fmind/agent-levers/agents-md
Clone the repo
git clone --depth 1 https://github.com/fmind/agent-levers

Made for: Codex, OpenCode.

Per session 630 This file is loaded in full into every session.
When invoked 630 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 $0.00630 $0.00630
Opus 5 $0.00315 $0.00315
Sonnet 5 $0.00126 $0.00126
Haiku 4.5 $0.00063 $0.00063

Measured yesterday against content hash cfb1cec22be3, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

agent-levers 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 yesterday.

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 · 37 lines

How it starts

The opening of the file, as written. The whole thing — 37 lines — stays where its author put it; the contents beside it link to each section on GitHub.

AGENTS.md

Overview

  • agent-levers ships a small set of slash-command skills (/lever-init, /lever-new, /lever, /lever-status) that give a coding agent a structured plan → do → check → act loop.
  • Skills are file-based and harness-agnostic: the same skills/ tree works in Claude Code, Gemini CLI, GitHub Copilot, and OpenCode.
  • Repo state is .agents/levers/<id>-<slug>/{lever.yaml, LEVER.md}lever.yaml validated against lever.schema.json.

Conventions

  • Line caps. Skill files exist to be loaded by an agent on every run; verbosity directly costs context budget. Targets:

    • SKILL.md — ≤ 110 lines
    • skills/lever/references/*.md — ≤ 100 lines (each step procedure)
    • skills/lever/templates/*.md — ≤ 50 lines (each LEVER.md section template)
    • skills/*/templates/*.md — ≤ 50 lines (other artifact templates)
    • Project-root markdown (README.md, AGENTS.md) — uncapped, but tighten when it sprawls

    When a file approaches its cap, prefer cutting motivational/explanatory prose over removing structural rules. Each rule should be stated once, in the imperative, in the file the agent loads when it needs it. The "why" lives in commit messages and PR descriptions.

  • Lint. Run npm run lint and npm run validate to validate changes — lint covers Prettier formatting and markdownlint-cli2; validate runs ajv against every examples/**/lever.yaml. Both gate pre-commit locally and CI.

  • Voice. Skill bodies address the agent in the imperative ("Read X, then write Y"). Avoid second-person flavor text aimed at the human reader — that belongs in the README.

Workflow

  • Use lever skills (this repo dogfoods its own framework on non-trivial changes):
    • /lever-new <title> — capture a new task (Session 1: discuss intent, write the brief)
    • /lever <id> — run the chain (Session 2: plan → do → check → act → done, autonomously until paused or finished)
    • /lever-status — inspect levers in flight; /lever-status <id> cancel [<reason>] — retire one

Read the full file on GitHub · 37 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. yesterday First seen · 37 lines · 630 tokens per session scan A cfb1cec22be3

Subscribe to this mod's changes

agent-levers AGENTS.md is an instructions file published in the GitHub repository fmind/agent-levers (3 stars, last pushed 3mo ago), licensed MIT. It adds 630 tokens to every session, about $0.0032 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.