| name | execute-sprint |
| description | Execute a sprint plan by reading task files, dispatching subagents per task, and managing review cycles. Uses subagent-driven-development pattern. |
Execute Sprint
Read a sprint plan, extract tasks, and dispatch subagents to implement each task with review cycles.
Usage
/execute-sprint <plan-file-path>
Example: /execute-sprint docs/superpowers/plans/2026-04-11-gap-closure-multi-sprint.md
Workflow
Step 1: Read the Plan
Read the specified plan file and extract:
- Sprint number and name
- Task list with dependencies
- Wave ordering (which tasks can run in parallel)
Step 2: Validate Prerequisites
Before executing:
- Ensure
npm run typecheck passes on current branch
- Confirm no uncommitted changes (
git status --porcelain)
- Create a feature branch if not already on one
Step 3: Execute Waves
For each wave (group of independent tasks):
- Dispatch agents — use the
superpowers:dispatching-parallel-agents skill to spawn one agent per task in the wave
- Wait for completion — all agents in the wave must finish before proceeding
- Review cycle — after each wave:
- Run
npm run lint && npm run typecheck
- Fix any integration issues (barrel exports, missing imports)
- Commit the wave's changes
Step 4: Between Waves
After each wave completes:
- Verify typecheck still passes
- Commit with message:
feat(<sprint>): complete wave <N> — <task summaries>
- Proceed to next wave
Step 5: Sprint Completion
After all waves complete:
- Run full verification:
npm run lint && npm run typecheck && npm run build
- Create summary of what was implemented
- Report any tasks that were blocked or deferred
Agent Dispatch Template
For each task, spawn an agent with:
Task: <task name>
Description: <from plan>
Files to modify: <from plan>
Acceptance criteria: <from plan>
Rules:
- Read CLAUDE.md before writing code
- Use @ui primitives — never raw HTML
- Run `npx eslint <changed-files>` and `npx tsc --noEmit` before reporting done
- Commit changes to your branch when complete
Error Handling
- If an agent fails, capture the error and continue with other agents in the wave
- After the wave, report failures and ask the user whether to retry or skip
- Never silently skip a failing task
Plan File Format
The plan should contain tasks in this structure:
### Task N: <name>
- **Wave:** <number>
- **Depends on:** <task numbers or "none">
- **Files:** <list of files to create/modify>
- **Description:** <what to implement>
- **Acceptance criteria:** <what "done" looks like>
