Sprint

Sprint Workflow Pipeline

A sprint is the full lifecycle of shipping a feature — from planning through deployment. This skill chains Ultraship skills into a structured pipeline where each phase produces artifacts that feed the next.

Announce at start: "I'm using the sprint workflow to guide this feature from plan to ship."

The Pipeline

Copy code
┌──────────┐    ┌──────────┐    ┌──────────┐    ┌──────────┐    ┌──────────┐    ┌──────────┐
│  PLAN    │───▶│  BUILD   │───▶│  TEST    │───▶│  REVIEW  │───▶│  SHIP    │───▶│  VERIFY  │
│          │    │          │    │          │    │          │    │          │    │          │
│ /write-  │    │ /execute-│    │ /tdd     │    │ /review  │    │ /deploy  │    │ /canary  │
│  plan    │    │  plan    │    │          │    │          │    │ /ship    │    │ /retro   │
└──────────┘    └──────────┘    └──────────┘    └──────────┘    └──────────┘    └──────────┘
     │                │               │               │               │               │
     ▼                ▼               ▼               ▼               ▼               ▼
  Plan doc       Working code    Tests pass     Issues fixed    Deployed         Verified

Phase 1: Plan

Trigger: User describes a feature, requirement, or bug to fix.

1. If the scope is large, use ultraship:brainstorming first to explore the idea space
2. Use ultraship:writing-plans to create an implementation plan
3. Plan is saved to docs/ultraship/plans/YYYY-MM-DD-<feature>.md

Artifacts produced: Implementation plan with file map, task list, test strategy

Gate: Plan must be reviewed and approved by user before proceeding to Build.

Phase 2: Build

Trigger: Plan is approved.

1. Use ultraship:executing-plans to implement the plan task by task
2. If tasks are independent, use ultraship:dispatching-parallel-agents for parallel execution
3. If working in isolation, use ultraship:using-git-worktrees for a clean workspace

Scaling to large work: Subagents can now nest up to 5 levels deep — a dispatched implementer can fan out its own helpers for sub-steps. For codebase-wide changes (a migration, a cross-cutting refactor, an audit of every route), prefer a dynamic Workflow: describe the task naturally and include the word "workflow" so Claude generates an orchestration script that pipelines the work across many background subagents with verification built in. Use a single controller for a handful of tasks; escalate to a Workflow when the work-list is large or unknown in size.

Artifacts produced: Working code, committed to a feature branch

Gate: All planned tasks are implemented. Code compiles/runs without errors.

Phase 3: Test

Trigger: Implementation is complete.

1. Use ultraship:test-driven-development to write tests for new code
2. Run the full test suite to catch regressions
3. If bugs are found, use ultraship:systematic-debugging to diagnose (not guess-and-fix)

Artifacts produced: Passing test suite, test coverage for new code

Gate: All tests pass. No known bugs in new code.

Phase 4: Review

Trigger: Tests pass.

1. Use ultraship:requesting-code-review for self-review
2. Run /review for automated code review with confidence scoring
3. Run /secure for security scanning
4. Fix any high-confidence issues found

Artifacts produced: Review report, security scan results, fixes committed

Gate: No critical or high-severity issues remaining.

Phase 5: Ship

Trigger: Review is clean.

1. Use ultraship:verification-before-completion for final verification
2. Run /ship for the full pre-deploy scorecard (SEO, security, code quality, bundle)
3. If score is acceptable, run /deploy for the deploy pipeline
4. Use ultraship:finishing-a-development-branch to merge/PR

Artifacts produced: Deploy to production, PR/merge to main

Gate: /ship scorecard is READY TO SHIP. Deploy succeeds.

Phase 6: Verify

Trigger: Deploy completes.

1. Run /canary to verify production health
2. If canary detects issues, escalate to /rescue
3. Save any deployment learnings via /learn
4. Run /retro at the end of the sprint to review overall progress

Artifacts produced: Canary report, learnings, retrospective

Workflow Rules

1. Never skip phases. Each phase exists because skipping it causes problems.
2. Gates are mandatory. Don't proceed to the next phase until the gate criteria are met.
3. Artifacts chain forward. Each phase's output is the next phase's input.
4. The user decides pace. Some sprints complete in an hour. Some take a week. The pipeline adapts.
5. Small batches. Prefer shipping small features frequently over large features infrequently.

Quick Sprint (for small changes)

For bug fixes or small features (< 50 lines of code), compress the pipeline:

1. Investigate → Use /investigate to find root cause
2. Fix + Test → Fix the bug, write a test
3. Verify → Run /review, check tests pass
4. Ship → Push and run /canary

Skip the full planning phase for changes that don't need it.

Sprint Status

Track sprint progress using tasks. At any point, the user can ask "where are we?" and get a clear answer:

Copy code
Sprint: Add webhook retry logic
Phase: 3/6 — TEST
Status: 2 tests written, 1 failing (timeout issue in retry delay)
Next: Fix failing test, then proceed to REVIEW