From ship
Routes ambiguous software-delivery requests to the smallest useful Ship workflow: a single skill, a phase bundle, or the full /ship:auto pipeline. Use when a user says build/check/ship without a phase.
How this skill is triggered — by the user, by Claude, or both
Slash command
/ship:use-shipThis skill is limited to the following tools:
The summary Claude sees in its skill listing — used to decide when to auto-load this skill
Use this skill as Ship's routing brain.
Use this skill as Ship's routing brain.
Your job is to choose the smallest useful Ship workflow for the request. Do not force the full pipeline unless the user explicitly wants production delivery or the risk justifies it.
/ship:auto workflow.Assume the host model is capable and literal:
| Need | Route |
|---|---|
| Understand scope, write a plan, de-risk approach | /ship:design |
| Architecture/API/data model decision | /ship:arch-design, then /ship:design if implementation planning is needed |
| Implement a well-scoped change | /ship:design → /ship:dev |
| Implement from an existing approved plan | /ship:dev |
| Add durable browser/API/CLI coverage | /ship:e2e |
| Check code correctness | /ship:review |
| Verify runtime behavior | /ship:qa |
| Harden a completed change | /ship:e2e → /ship:review → /ship:qa → /ship:refactor |
| Prepare delivery after work is complete | /ship:handoff |
| End-to-end production delivery | /ship:auto |
Ship's auto runner does not prescribe per-stage YAML schemas.
.ship/tasks/<task_id>/input/requirement.md when a full
flow is running..ship/tasks/<task_id>/control/ if
it helps the current task, but the agent chooses the shape.When a task needs durable product, design, engineering, quality, delivery, or archive artifacts, use the repository's existing convention first. If there is no convention, create the smallest useful folder under:
docs/ship/<task-id-or-req-id>/
input/ # raw requirement and source notes
product/ # clarified requirements and acceptance criteria
design/ # UI/product design artifacts when needed
engineering/ # architecture and implementation plans
quality/ # test plans, QA reports, quality evidence
delivery/ # handoff, release notes, rollout notes
archive/ # final summary and decisions
Only create the folders that the task needs. Prefer Markdown. Use YAML or JSON only when a later agent, script, or check will consume the structure. Keep code in the real repository tree; this folder is for durable production artifacts.
Atomic skills must remain usable without /ship:auto or any task directory. If
the user names a phase directly, run that phase directly.
When intent is ambiguous, route to a bounded bundle instead of asking for the full pipeline by default. Examples:
/ship:design/ship:design → /ship:dev/ship:review and/or /ship:qa/ship:autodocs/ship/<task-id-or-req-id>/ artifacts, using focused skills
such as /ship:write-docs, /ship:arch-design, or /ship:visual-designnpx claudepluginhub heliohq/ship --plugin shipProvides Shipyard overview and slash commands for Claude Code projects: init, brainstorm, plan phases, build with reviews/audits, quick tasks, ship, and skill access.
Universal shipping workflow: 8-phase linear pipeline from verification to deploy. Reads type-checklists.md to select the right checklist for the artifact type. The ONLY pause is Phase 7 (user confirmation before irreversible deploy/publish). TRIGGER when: user wants to ship, release, publish, or deploy something; user asks "is this ready to ship?"; user wants a pre-release checklist; user wants to publish a package, CLI, API, web app, ML model, skill, docs site, infrastructure, or research paper; user invokes /autoresearch:ship. DO NOT TRIGGER when: user wants iterative development (use autoresearch); user wants to debug errors (use autoresearch:fix); user just wants a code review.