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/heliohq/ship/agents-mdgit clone --depth 1 https://github.com/heliohq/shipWrote 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/heliohq/ship/agents-md)<a href="https://agentmods.dev/instructions/heliohq/ship/agents-md"><img src="https://agentmods.dev/badge/instructions/heliohq/ship/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 | $0.01117 | $0.01117 |
| Opus 5 | $0.00558 | $0.00558 |
| Sonnet 5 | $0.00223 | $0.00223 |
| Haiku 4.5 | $0.00112 | $0.00112 |
Grade A, and why
ship 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 — 69 lines — stays where its author put it; the contents beside it link to each section on GitHub.
AGENTS.md
Commands
| Action | Command |
|---|---|
| Validate JSON | jq . <file> |
| Lint shell | shellcheck scripts/*.sh (if installed) |
| Test hooks | echo '<json>' | bash scripts/<hook>.sh |
| Reload plugin | /reload-plugins in Claude Code |
Repository Map
| Directory | Contents | Purpose |
|---|---|---|
scripts/ |
Shell scripts | Startup hint, workflow hooks, auto orchestrator, PR readiness, docs index generation, PATH bootstrap, and dev-phase file handoffs (story-brief.sh, review-package.sh) |
hooks/ |
hooks.json, codex-hooks.json |
Same three hooks per platform (session hint, phase guardrail, stop gate); they differ only in the root variable each platform expands |
.codex-plugin/ |
plugin.json |
Native Codex plugin metadata for skills, MCP, and Codex UI |
skills/ |
Skill dirs plus shared/ helpers |
Slash commands (/ship:*) and shared references (shared/ has no SKILL.md, so it is not a routable skill) |
skills/use-ship/ |
Routing skill | Agent-facing guide for grouping Ship phases based on task need |
skills/auto/prompts/ |
.md.tmpl files |
Prompt templates for the full workflow runner |
.claude-plugin/ |
plugin.json |
Plugin metadata for ShipAI |
.mcp.json |
MCP config | Codex MCP server registration |
Architecture
Ship has one opt-in workflow layer. It fires only during user-triggered Ship
workflows, especially /ship:auto:
session-start.shinjects only a small/ship:use-shiprouting hint; no docs index or artifact content./ship:use-shipchooses the smallest useful standalone skill, phase bundle, or full workflow.- Full workflow runs write raw input to
.ship/tasks/<task_id>/input/requirement.md. - The runner owns exactly one state surface: the frontmatter of
.ship/ship-auto.local.md. There is no side-state; agents may create task-specific notes under the task dir if useful. - Markdown artifacts and repository code are the output plane.
- Bulk artifacts move between agents as file paths, never pasted prompt text: story briefs, implementer reports, and review packages live in the self-ignoring
.ship/scratch/. - Durable production artifacts may be organized under
docs/ship/<task-id-or-req-id>/when the repo lacks an existing convention. stop-gate.sh— blocks session exit while.ship/ship-auto.local.mdis active (with fast-path for terminal phases)auto-orchestrate.sh— code-driven state machine for staged workflows (init, resume, complete, status commands)phase-guardrail.sh— PreToolUse hook enforcing artifact access rules per phase (QA independence, review read-only, state file protection)- Claude Code loads hooks through plugin
hooks/hooks.json(${CLAUDE_PLUGIN_ROOT}). Codex loadshooks/codex-hooks.json(${PLUGIN_ROOT}) via the explicit"hooks"pointer in.codex-plugin/plugin.json— the pointer is REQUIRED: without it Codex falls back tohooks/hooks.json, whose variable does not expand there. Both SessionStart hooks matchstartup|clear|compactso the hint does not re-inject on resume
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 · 69 lines · 1,117 tokens per session scan A 6f190bb5e9a1
ship AGENTS.md is an instructions file published in the GitHub repository heliohq/ship (92 stars, last pushed 2mo ago), licensed MIT. It adds 1,117 tokens to every session, about $0.0056 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.
Other instructions, from other repositories
marketing-machine CLAUDE.md
Instructions for cagatayuncu/marketing-machine, covering marketing-machine — contributor notes, commands, architecture in ten lines and invariants (violating these is a bug, not a style choice).
harness-evolver CLAUDE.md
Instructions for raphaelchristi/harness-evolver, covering claude.md, what this is, dependencies, running tools locally and use --no-canary to skip preflight check.
agent-skills GEMINI.md
Gemini CLI instructions for motherduckdb/agent-skills, covering motherduck skills for gemini cli, default routing, skill catalog, utility and workflow.
digital-brain copilot-instructions.md
Copilot instructions for eliransu/digital-brain, covering digital-brain: github copilot instructions, project type, repository layout, conventions copilot should follow and when editing skills (skills/ /skill.md).
vscode buildNext.instructions.md
Working notes and architecture documentation for the new esbuild-based build system in build/next. Use when making changes to the new build pipeline (transpile/bundle commands, NLS plugin, source-map handling, resource copying, or self-hosting watch tasks).
spec-kit AGENTS.md
AGENTS.md instructions for github/spec-kit, covering agents.md, about spec kit and specify, quickstart — add a new integration in 5 steps, integration architecture and integrationmanifest — file tracking.