From dream-team
Pick relevant Dream Team agents to collaboratively plan a feature
How this command is triggered — by the user, by Claude, or both
Slash command
/dream-team:dream-plan Feature to plan (e.g., 'user notifications system')The summary Claude sees in its command listing — used to decide when to auto-load this command
# Dream Plan: $ARGUMENTS You are planning the feature described above using the Dream Team methodology, but with **selective agent activation** — only launch agents that are relevant to this specific feature. ## Agent Selection Read `references/agent-roster.md` for the full roster with preferred agents, fallback prompts, and inclusion criteria. **Planning uses all agents including Code Quality Engineer** (the only phase where it's active — `/simplify` replaces it during implementation and review). Review the feature and decide which agents to activate. **Justify each inclusion/exclusion...
You are planning the feature described above using the Dream Team methodology, but with selective agent activation — only launch agents that are relevant to this specific feature.
Read references/agent-roster.md for the full roster with preferred agents, fallback prompts, and inclusion criteria. Planning uses all agents including Code Quality Engineer (the only phase where it's active — /simplify replaces it during implementation and review).
Review the feature and decide which agents to activate. Justify each inclusion/exclusion in a brief sentence before launching. For trivial changes (single-line fixes, typo corrections), skip the full dream-team process and just make the change directly.
Select agents: State which agents you're activating and why (1-2 sentences each). State which you're skipping and why.
Launch in parallel: Launch all selected agents simultaneously using the Task tool with run_in_background: true. Each agent receives the full feature description and is told to focus on planning only — no implementation.
Each agent prompt must include:
Synthesize: After all agents return, combine their findings into a unified plan. Highlight:
Devil's Advocate pass: Resume the Devil's Advocate agent with the synthesis for a final challenge. Address every objection — either incorporate it or explain why it's dismissed.
Edge case sweep: Review the plan yourself with fresh eyes and specifically hunt for missed edge cases. For each component in the plan, ask:
Add any newly discovered edge cases to the plan. If an edge case would require a new task, add it. If it's a risk, add it to the risks table.
Create sprint directory: Save the plan to the project's planning directory:
planning/sprints/sprint-N-name/
├── plan.md — Copy from templates/sprint-plan.md, fill in architecture decisions
├── tasks.md — Copy from templates/tasks.md, fill in implementation tasks
├── progress.md — Copy from templates/progress.md, initialize checkpoint
└── test-plan.md — What to test (if Test Engineer was activated)
Use the templates from this plugin's templates/ directory as starting points.
Final plan: Present the consolidated implementation plan with:
Present the final plan clearly and ask the user if they'd like to proceed with implementation.
Read references/sprint-rules.md for the full non-negotiable sprint rules. Key planning-specific rules:
planning/sprints/sprint-N-name/test-plan.md during planning if the sprint has testable featuresnpx claudepluginhub numanumanuma/claude-plugins --plugin dream-team/planProduces a step-by-step technical implementation plan from an approved feature spec, identifying files, functions, and data flows without writing code.
/plannewCreates new development plan from scratch using feature/project description. Enters plan mode, enforces TDD/DRY/SOLID principles, sets up tasks with strict rules.
/planGenerates comprehensive implementation plan: restates requirements, identifies risks and dependencies, outlines phases, estimates complexity, and waits for user confirmation.
/planGenerates detailed 3-phase implementation plans covering requirements, design, tasks, risks, and quality checks for any project. Awaits explicit user approval before coding.
/planCreates a step-by-step implementation plan from requirements or a PRD markdown file, then waits for user confirmation before any code changes.
/planRestates requirements, assesses risks, creates step-by-step implementation plan with phases, dependencies, complexity estimates, and waits for user confirmation.