An EverydayAI fork of obra/superpowers
Adds a full SDLC orchestration pipeline on top of the Superpowers skills system for Claude Code.
Pipeline v3 turns Claude Code into an autonomous software development organization. Five departments — Product, Design, Engineering, Quality, and Marketing — run in parallel within token-budgeted iterations, coordinated by an Orchestrator agent and governed by a Founder agent at Integration Points.
The pipeline takes a product idea from discovery through shipping code:
PI Planning → Iteration 1 (Foundation) → Iteration 2 (Build) → Iteration 3 (Harden) → Ship
Each iteration runs all active department teams concurrently. Integration Points fire between iterations for cross-department sync and Founder decisions.
| Command | Description |
|---|---|
/orchestration |
Run the full Pipeline v3 SDLC — PI Planning through ship |
/scaffold |
Generate project scaffolding from YAML templates |
/pm-discover |
Standalone PM discovery (personas, journeys, stories, PRD, spec) |
Founder (opus) — final decision maker
│
├── Product Department
│ ├── Product Lead (opus)
│ ├── Domain Researcher · Market Researcher · UX Researcher
│ ├── Onboarding Designer · Artifact Generator
│
├── Design Department
│ ├── Design Lead (opus)
│ └── Visual Designer
│
├── Engineering Department
│ ├── Engineering Lead (opus)
│ ├── Implementers (subagent-driven) · Code Reviewers
│ └── UX Reviewer
│
├── Quality Department
│ ├── Quality Lead (opus)
│ ├── Verifier · Browser Tester · Security Reviewer
│ └── Release Readiness Reviewer
│
├── Marketing Department (optional)
│ ├── Marketing Lead (opus)
│ ├── Content Marketer · Growth Marketer · SEO Specialist
│
└── Staff
├── Orchestrator (opus)
└── Humanizer
- Parallel iterations — all teams work concurrently within each iteration
- Token-budgeted timeboxes — teams get token budgets enforced by hooks
- Dual-trigger Integration Points — syncs fire when artifacts exist OR budgets exhaust
- Hook-automated coordination — 6 shell hooks handle state management at zero orchestrator token cost
- Structured memory — cross-project learnings compound via
knowledge-compoundingskill - SAFe alignment — Epic/Capability/Feature/Story hierarchy, WSJF prioritization, PI Planning
All original Superpowers skills remain available. The fork extends but does not modify the upstream skills system.
Testing
- test-driven-development — RED-GREEN-REFACTOR cycle
Debugging
- systematic-debugging — 4-phase root cause process
- verification-before-completion — Ensure it's actually fixed
Collaboration
- brainstorming — Socratic design refinement
- writing-plans — Detailed implementation plans
- executing-plans — Batch execution with checkpoints
- dispatching-parallel-agents — Concurrent subagent workflows
- requesting-code-review / receiving-code-review — Code review workflow
- using-git-worktrees — Parallel development branches
- finishing-a-development-branch — Merge/PR decision workflow
- subagent-driven-development — Fast iteration with two-stage review
Meta
- writing-skills — Create new skills following best practices
- using-superpowers — Introduction to the skills system
This fork is designed for use as a Claude Code plugin. See the upstream obra/superpowers README for standard installation instructions.
- Judgment before tools — severity gates surface decisions that matter
- Research before ask — exhaust automated research before asking the human
- Verify before trust — anti-rationalization verification at every stage
- Don't reinvent — evaluate existing solutions before building custom
- Leads resist — department Leads push back on mediocre work
- Compound intelligence — every pipeline run makes the next one smarter
Built on top of Superpowers by Jesse Vincent (obra). If Superpowers has helped you, consider sponsoring Jesse's open source work.
MIT License — see LICENSE file for details.