superpowers: A complete software development methodology for coding agents, built from composable skills and initial instructions.
What problem it solves
Coding agents can begin writing code before clarifying the intended outcome, design, and implementation plan. Superpowers addresses this by guiding agents through structured brainstorming, planning, implementation, testing, review, and branch completion workflows.
Key features
Composable skills — The framework provides skills for testing, debugging, collaboration, planning, execution, review, and skill authoring. Relevant skills are selected automatically before tasks begin.
Source: README, skills
Structured development workflow — Agents clarify intent, obtain design approval, create detailed plans, work in isolated branches, implement tasks, and select how to finish the branch.
Source: README
Plan-driven implementation — Plans divide work into small tasks with exact file paths, complete code, and verification steps. Teams can choose between subagent-driven development with per-task review and inline plan execution with a final review.
Source: README
Testing and verification — The workflow enforces RED-GREEN-REFACTOR during implementation and includes systematic debugging and verification skills for diagnosing failures and confirming that issues are fixed.
Source: README
Multi-harness support — The framework provides installation instructions for several coding-agent environments, with separate installation required for each harness.
Source: README, .claude-plugin, .codex-plugin
Quick start
Copied verbatim from the project's README.
/plugin install superpowers@claude-plugins-officialagy plugin install https://github.com/obra/superpowers devin plugins install obra/superpowersWho it may suit
This may suit developers who want coding agents to follow a structured path from requirements through testing and review, and teams using one or more of the supported agent harnesses.
Topics to discuss
Which parts of the brainstorming-to-review workflow should be mandatory for different kinds of projects?
How should additional skills be designed and tested so they work across every supported coding agent?
Which signals should agents use to decide between subagent-driven development and inline plan execution?
Written by ReadmeX AI from the project's README, source and public GitHub data; it may contain mistakes.
Read the original README https://github.com/obra/superpowers#readme
ReadmeX opened this community for a trending project; it is not run by the project. Maintainers can claim it from the community page.
Comments
…