agentic-ship AGENTS.md

A set of instructions for working with the Agentic Ship repository, a tool-only project that provides guidance, skills, and agent definitions rather than a finished application.

In plain words
What is it for?
Use it when an agent needs to understand the project’s commands, skills, agent files, and configuration for downstream product work.
Why use it?
It explains the repository structure and keeps different coding agents following the same rules.

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/moasq/agentic-ship/agents-md
Clone the repo
git clone --depth 1 https://github.com/moasq/agentic-ship

Made for: Codex, OpenCode.

Per session 11,608 This file is loaded in full into every session.
When invoked 11,608 The same file — it is already loaded in full.
Security scan C 1 finding. 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 $0.11608 $0.11608
Opus 5 $0.05804 $0.05804
Sonnet 5 $0.02322 $0.02322
Haiku 4.5 $0.01161 $0.01161

Measured 2d ago against content hash 38bcd2a13f76, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade C, and why

agentic-ship AGENTS.md scanned grade C with 1 finding 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 2d 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.

Recursive force deletehighDestructive command

rm -rf with a variable or a broad path is one typo away from removing the wrong tree.

**Never write `cp`, `ln`, `readlink`, `grep`, `rm -rf`, `mkdir -p`, `chmod`, `openssl`,
AGENTS.md · 726 lines

How it starts

The opening of the file, as written. The whole thing — 726 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Agentic Ship

This is the tool-only repository. It has no bundled application, backend, assets, or deployment. The product-stack rules below are guidance for a downstream product workspace that adopts Agentic Ship; do not expect src/, convex/, or a local website to exist here.

This file is where every rule is declared. CLAUDE.md imports it with one line; .claude/skills links to .agents/skills. Skills elaborate these rules into procedure — they must never restate one differently, and never introduce a rule that is not declared here. When a skill and this file disagree, this file is right and the skill is a bug.

Works with any agentic tool. Codex, Cursor, Windsurf, Cline, Copilot, Gemini CLI, and OpenClaw (with this repository as its agent workspace) read this file natively; skills are plain markdown any agent can follow. Canonical role briefs live in .agents/agents/; pnpm sync:agents generates the Claude plugin's top-level agents/, project-native Codex and Cursor agents, and non-secret Hermes and OpenClaw profiles. Cursor gets MCP through the committed .cursor/mcp.json mirror; Codex gets project-scoped TOML generated from the same pinned catalog. The repo is also an installable plugin for Claude Code and Codex — manifests in .claude-plugin/ and .codex-plugin/ point at the same .agents/skills/, so the plugin adds a second delivery path, never a second copy of a rule. Per-tool matrix and sync rules: .agents/skills/agent-compatibility/SKILL.md. Skills with a references/ folder keep their deep material there — load it only when the task needs it.

Instructions live here. Procedures live in .agents/skills/. Tool wiring lives in .mcp.json. Product, feature, and human-input contracts live in .agents/contracts/. Safe runtime coordination lives under gitignored .agent-state/; credentials never do. The project-scoped agentic-ship MCP server wraps the real health, verification, connection, queue, and UI-evidence services. Its mutation tools require the explicit --allow-mutations server argument; its full contract lives in .agents/skills/agent-compatibility/references/agentic-ship-mcp.md. Plugin wiring lives in .claude/settings.json (the official nextjs plugin from the vercel/next.js repo is declared there). Provenance for all of it lives in skills.lock.json. One rule, one home: it is declared here, applied there.

Read the full file on GitHub · 726 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. 2d ago First seen · 726 lines · 11,608 tokens per session scan C 38bcd2a13f76

Subscribe to this mod's changes

agentic-ship AGENTS.md is an instructions file published in the GitHub repository moasq/agentic-ship (10 stars, last pushed 4d ago), licensed MIT. It adds 11,608 tokens to every session, about $0.0580 per session on Opus 5. A static security scan graded it C with 1 finding (recursive force delete). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.