Installs into .claude/skills of the current project.
Sponsors
SSkills DirectorySkills Directory
Know which skills are safe — weekly.
SSkills DirectorySkills Directory
Is that skill safe to install?
Are you the author of Ship Sprint?
Add the live security badge to your README. It updates with every re-scan.
[](https://www.skillsdirectory.com/skills/acendas-ship-sprint)
---
name: ship-sprint
description: "Plan a new sprint or cancel an active one."
allowed-tools: [Read, Write, Edit, Grep, Glob, LSP, Agent, AskUserQuestion, WebSearch, WebFetch, TaskCreate, TaskUpdate, TaskList, "Bash(shipyard-context:*)", "Bash(shipyard-data:*)"]
model: opus
effort: low
argument-hint: "[--cancel] [--hours <n> | --until <time>]"
---
# Shipyard: Sprint Planning
Plan a new sprint by pulling features from the backlog and decomposing into waves.
## Context
!`shipyard-context sprint-planning`
**Shell.** Shipyard binaries you may invoke from Bash: `shipyard-data archive-sprint <id>`, `shipyard-data metrics ...`, `shipyard-data init-sprint <id>` (Step 12), `shipyard-data draft set-sprint-status ...`, `shipyard-data sprint set <key> <value>` and `shipyard-data sprint check` (Step 12). Run commands bare from the project root. SPRINT.md and PROGRESS.md are created by the `init-sprint` CLI; **SPRINT.md frontmatter is thereafter mutated only via `shipyard-data sprint set`, never a hand Edit** (hand-editing frontmatter is the corruption class that welds keys together). **PROGRESS.md and the pipeline cursors are CLI/render-owned — the model never writes them** (a PreToolUse hook denies such writes). Metrics are CLI-owned: `memory/metrics.json` is authoritative and `memory/metrics.md` is generated; never Write/Edit either metrics file by hand. Use the Write tool for arbitrary *narrative* artifacts (story files, step files, SPRINT-DRAFT.md); the SPRINT.md *body* (Goal, `### Wave N`, Critical Path, Risks, Swap Log) stays a model Edit on the CLI-created file.
**Render before asking.** Follow SHIPYARD_STARTUP_RULES and SHIPYARD_ONBOARDING_INSTRUCTION from startup.
**Quiet by default.** Between user-input gates work quietly; render feature lists, wave breakdowns, and quality-gate results in full only at a gate (render-before-ask). **No running commentary** ("Now I'll…", "Let me…"). Doctrine: `${CLAUDE_PLUGIN_ROOT}/skills/ship-discuss/references/communication-design.md` § "Interim Communication: Quiet by Default".
**Capability-skill playbooks.** Where a step says *"follow the `X` playbook"* or "dispatch `X`", X is a capability skill — **Read** `${CLAUDE_PLUGIN_ROOT}/skills/<X>/SKILL.md` and execute it inline; never hand it to the `Skill` tool (capability skills are `disable-model-invocation: true`, so `Skill` refuses them).
## Input
$ARGUMENTS
## Session Mutex Check
**Absolute first context action — before mode detection, before any manual reads.** The `shipyard-context sprint-planning` pre-exec block above acquires the planning lock and returns the planning context bundle in one CLI call. If that context contains `SHIPYARD_LOCK_ACQUIRED=false`, echo the `SHIPYARD_LOCK_BLOCKED:` `⛔` block text verbatim as the entire response and STOP — do not continue with any other instructions, do not call any other tools. If it contains `SHIPYARD_LOCK_NOTICE:`, echo that stale/corrupt-lock recovery line once, then proceed. Do not run a separate lock-acquire call; the bundled context command already did it (see the `acquiring-skill-lock` capability skill for the full contract — locks are CLI-owned; never Read or Write either lock file by hand).
## Detect Mode
- If `--cancel` → CANCEL mode
- `--hours <n>` / `--until <ISO time | HH:MM>` → an optional TIME budget for PLAN mode (points stay the default). Carry the flag verbatim into Step 1's `shipyard-data capacity` call; never convert hours to points yourself. Both flags, or a time flag plus a points answer, is a usage error the CLI reports — relay it and ask the user to pick one.
- If the context has `SHIPYARD_NEXT_SPRINT_READY=<id>` → an approved sprint is waiting and nothing is running: **Read and follow** `${CLAUDE_PLUGIN_ROOT}/skills/ship-sprint/references/draft-ahead.md` § "Promote".
- If the context has `SHIPYARD_SPRINT_PLAN_MODE=draft-ahead` (a sprint is running or waiting for its review; planning is never blocked by it):
Read current sprint state — stories, steps per wave, progress, remaining capacity. Write one short summary in chat, then ask one question:
- **Draft the next sprint (Recommended)** → PLAN mode, drafted ahead: **Read and follow** `references/draft-ahead.md` § "Draft ahead" on top of PLAN mode. The running sprint is not touched.
- **Add stories to the running sprint** → EXTEND mode. Offer only when `SHIPYARD_RUNNING_BESIDE` names neither `ship-execute` nor `ship-review`.
- **Cancel it and replan** → CANCEL mode, then PLAN mode. Same condition.
- **Wait** → stop; suggest `/ship-execute` or `/ship-review`.
- If completed sprint exists in `current/` (status: `completed` but not archived):
Show what was found:
```
Found completed sprint [ID] that wasn't archived:
Features: [list with statuses]
Branch: [branch name] — [merged/unmerged]
Velocity recorded: [yes/no]
```
Handle cleanup transparently:
1. Archive by running `shipyard-data archive-sprint sprint-NNN` from Bash (substitute the real sprint ID). This atomically records velocity from delivered features, renames `<SHIPYARD_DATA>/sprints/current/` → `<SHIPYARD_DATA>/sprints/sprint-NNN/`, and recreates an empty `current/`. Do NOT synthesize raw `cp`/`mv`/`mkdir` commands against the plugin data dir — they're not portable and not atomic.
<!-- AC-6 -->
If `archive-sprint` exits non-zero (the CLI refuses a completed sprint that `/ship-review` never processed), show the CLI's reason, tell the user review is pending, and AskUserQuestion: **Run /ship-review first (Recommended)** → stop and tell the user to run `/ship-review`; **Archive anyway** → run `shipyard-data archive-sprint sprint-NNN --unreviewed`.
2. Report: "Archived sprint [ID]. Velocity recorded for the next sprint."
3. Then proceed to PLAN mode
- Otherwise → PLAN mode
---
## PLAN Mode
**Read** `${CLAUDE_PLUGIN_ROOT}/skills/ship-sprint/references/plan-mode.md` and `${CLAUDE_PLUGIN_ROOT}/skills/ship-sprint/references/plan-shared-steps.md` then follow.
**Read** `references/quality-manifest.md`; on-demand (inline fallback, final step-file write): `references/task-decomposition-patterns.md`, `references/task-kinds.md`, `references/task-tech-notes-template.md`. Also: `references/carry-over-scan.md`, `references/implementation-decisions.md`, `references/planning-checklists.md`, `references/spec-validation.md`, `references/sprint-analyst-report.md`, `references/wave-decomposition.md`.
---
## EXTEND Mode (add features to active sprint)
Add features to an in-progress sprint without cancelling.
**Read** `${CLAUDE_PLUGIN_ROOT}/skills/ship-sprint/references/extend-mode.md` and `${CLAUDE_PLUGIN_ROOT}/skills/ship-sprint/references/plan-shared-steps.md` then follow.
---
## CANCEL Mode
**Read** `${CLAUDE_PLUGIN_ROOT}/skills/ship-sprint/references/cancel-mode.md` and follow it.