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/bonnguyenitc/specship/claude-mdgit clone --depth 1 https://github.com/bonnguyenitc/specshipWhat 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 | $0.00643 | $0.00643 |
| Opus 5 | $0.00321 | $0.00321 |
| Sonnet 5 | $0.00129 | $0.00129 |
| Haiku 4.5 | $0.00064 | $0.00064 |
Grade A, and why
specship CLAUDE.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.
How it starts
The opening of the file, as written. The whole thing — 44 lines — stays where its author put it; the contents beside it link to each section on GitHub.
specship — dev guide
npm package that installs a spec→plan→coding→review skills workflow into
projects (Claude Code, Codex, Cursor, Antigravity). Entry: bin/cli.js →
src/cli.js; install logic in src/init.js + src/targets.js; skill sources
in skills/; subagent definitions in agents/ (installed only for targets
declaring subagents in targets.js — today just Claude Code); per-agent
templates in .claude/, .codex/, .cursor/, .antigravity/.
Naming: ss- prefixes skills, never task state. Every skill folder and
frontmatter name: is ss-<skill> (skills/ss-spec/, /ss-ship), as are the
subagents in agents/ (ss-explorer, …). The pipeline's state vocabulary
stays bare: stage:, artifacts:, next_phase:, resume_phase:, --phase,
and the Pipeline Log's stage labels are all spec|plan|coding|review|debug|done.
Prefixing those would break every existing task.md. STAGE_SKILLS in
src/init.js is keyed by directory name, so it carries the prefix.
Verify every change
Always prove a change works before declaring it done:
npm test— zero-dep test suite (test/pipeline.test.js+test/cli.test.js). Fixtures must describe tasks that could actually exist: atask.mdwhoseartifacts:map names aconfirmedspec needsspec.mdon disk to match (use thefilesForhelper). Fictional fixtures hid a whole class of bugs once — seetasks/LESSONS.mdL2/L3.- Real install:
node bin/cli.js init --all --dir "$(mktemp -d)", then inspect the resulting tree (marker block in CLAUDE.md, skills copied, codex-onlyopenai.yamlmanifest, claude-only.claude/agents/subagents and nowhere else). - Anything touching packaging or
publish.sh:./publish.sh --dry-run.
Learnings
When Claude makes a mistake in this repo, don't just correct it in chat — add the lesson here as a bullet so it never repeats.
.claude/CLAUDE.mdis not only this repo's project instructions: it is the templatespecship initmerges into consumers' CLAUDE.md. Edits there ship to users — dev-only guidance belongs in this file (root CLAUDE.md).- package.json
filesmust list.claude/CLAUDE.md(the template), not the whole.claude/dir, so local dev config (settings.json,commands/) never ships in the npm tarball.
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.
- yesterday First seen · 44 lines · 643 tokens per session scan A e5812e593ec5
specship CLAUDE.md is an instructions file published in the GitHub repository bonnguyenitc/specship (2 stars, last pushed 23d ago), licensed MIT. It adds 643 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.
Other instructions, from other repositories
open-swe AGENTS.md
Instructions for langchain-ai/open-swe, covering agents.md, project, commands, architecture and entrypoints.
open-swe CLAUDE.md
Instructions for langchain-ai/open-swe, covering claude.md, project, commands, architecture and entrypoints.
cyrus copilot-instructions.md
Copilot instructions for cyrusagents/cyrus: Note, there is a need to maintain the use of '--print' when running the claude exec commands because that is what makes it non-interactive.
open-vibe-island AGENTS.md
AGENTS.md instructions for Octane0411/open-vibe-island, covering agents, goal, required workflow, commit policy and safety rules.
open-vibe-island CLAUDE.md
Claude Code instructions for Octane0411/open-vibe-island, covering claude.md, project, architecture, build & run and dev app (open island dev.app).
lastlight CLAUDE.md
Instructions for nearform/lastlight, covering last light — monorepo orientation, workspace map, published packages (seven), dependency graph (workspace edges) and logging — use the logger, never console.