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 commands/dwarvesf/dwarves-kit/shipgit clone --depth 1 https://github.com/dwarvesf/dwarves-kitWhat 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 | $0.00028 | $0.03679 |
| Opus 5 | $0.00014 | $0.01840 |
| Sonnet 5 | $0.00006 | $0.00736 |
| Haiku 4.5 | $0.00003 | $0.00368 |
Grade A, and why
ship 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 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.
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 — 216 lines — stays where its author put it; the contents beside it link to each section on GitHub.
You are a release engineer. The user says the feature is done. Your job is to verify, package, and ship it cleanly.
Process
Step 1: Review gate
Check if a review has been done:
- Resolve the active spec (
docs/specs/SPEC-NNN-<slug>.md, the SPEC-005 rule) and read its## Reviewsection verdict. - If verdict is
SHIPorFIX THEN SHIP(with fixes applied): proceed. - If verdict is
DO NOT SHIP: STOP. Tell the user to fix the issues first. - If the spec has no
## Reviewsection: WARN. Ask: "(A) Run /kit:review first / (B) Run /kit:review-team for thorough review / (C) Skip review and ship anyway"
Do not silently skip the review check. The user must explicitly choose to ship without review.
Step 1b: Completeness log (warn, not block)
Read ~/.claude/dwarves-kit/logs/completeness.log (the warn+log sink from the WORKFLOW completeness clauses). Surface any entries since the last ship/tag: lost build-decisions (decision-translation) and un-updated companion docs (doc-update, per the WORKFLOW doc-impact map). REPORT them so the maintainer decides; do NOT auto-block on completeness. Hard blocks stay reserved for the spec's ## Review DO-NOT-SHIP verdict and the safety gates. If the log is absent or empty, say "completeness: clean". Source: SPEC-006.
Step 1c: End-user guide check (warn, not block)
Apply the test: does this change have an end user who is not the builder? A library, an internal tool, or infra with no such user is exempt.
If yes: confirm GUIDE.md exists at the product root and still matches what
shipped (what it does, how to use it, what to do when it breaks). Template:
docs/GUIDE.template.md.
If GUIDE.md is missing or stale, REPORT it to the maintainer; do NOT block
the ship. Same voice as Step 1b / Step 4a: hard blocks stay reserved for the
spec's ## Review DO-NOT-SHIP verdict and the safety gates. Source: SPEC-216.
Step 2: Run tests
Detect the test runner and execute:
- Node.js:
npm testorpnpm testoryarn test - Go:
go test ./... - Python:
pytestorpython -m pytest - Rust:
cargo test
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.
- 2d ago First seen · 216 lines · 28 tokens per session scan A df46fe4c1477
ship is a command published in the GitHub repository dwarvesf/dwarves-kit (11 stars, last pushed 2d ago), licensed MIT. It adds 28 tokens to every session and 3,679 once invoked, about $0.0001 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 commands, from other repositories
pr-submit
Create git commit and submit current branch with Graphite (squashes commits and rebases stack).
git-pr-push
Create git commit and push branch as PR using git + GitHub CLI.
OPSX: Archive
Archive a completed change in the experimental workflow.
commit
智能 Git 提交：分析改动生成 Conventional Commit 信息，支持拆分建议.
ship
Commit all changes, push branch, create/update PR, and run local validation.
speckit.aftercode
Automatically execute the subsequent code delivery pipeline: skilltypesync → changelog → commit → prgenerate, in strict sequential order.