Superpowers is a set of composable skills plus bootstrap instructions that give a coding agent a fixed software development process. The skills trigger automatically: before any task, the agent checks which skill applies, and the workflows are mandatory rather than suggestions.
When you start building something, the agent first asks what you are trying to achieve, then presents the resulting design in short sections for sign-off. From the approved design it writes an implementation plan of small tasks, each with exact file paths, complete code, and verification steps, following red/green test-driven development, YAGNI (You Aren't Gonna Need It), and DRY. On your go-ahead, it either dispatches a fresh subagent per task with a review after each, or executes the plan inline with one review of the whole branch at the end.
The same skills install as a plugin or extension into many coding agents, including Claude Code, Codex, Cursor, Gemini CLI, GitHub Copilot CLI, OpenCode, Pi, Qwen Code, Hermes, Kimi Code, and Antigravity. A diagnosing skill reads a session transcript to explain why a skill fired or stayed silent, and can package a scrubbed bug-report bundle.
Features
- Brainstorming: refines a rough idea through questions, explores alternatives, and presents the design in sections for validation
- Git worktrees: creates an isolated workspace on a new branch and verifies a clean test baseline after design approval
- Plan writing: breaks work into tasks of a few minutes each, with file paths, complete code, and verification steps
- Subagent-driven development: dispatches a fresh subagent per task with two-stage review for spec compliance and code quality
- Inline plan execution: runs every task in the current session with one final review of the branch
- Test-driven development: enforces the red-green-refactor cycle and deletes code written before its tests
- Systematic debugging: a four-phase root-cause process with tracing and defense-in-depth techniques
- Code review: requesting and receiving review between tasks, with critical issues blocking progress
- Branch finishing: verifies tests and offers merge, pull request, keep, or discard, then cleans up the worktree
- Session diagnosis: reads a session transcript and reports what went wrong with line-level evidence
- Skill authoring: a skill for writing and testing new skills
