Superpowers

Superpowers is an agentic skills framework that transforms how AI coding agents build software. Instead of letting your coding agent jump straight into writing code, Superpowers forces it to step back, ask what you are really trying to do, and tease out a proper specification before writing a single line. Once you sign off on the design, Superpowers has your agent create an implementation plan broken into bite-sized tasks, each with exact file paths, complete code, and verification steps. Then it launches subagent-driven development: fresh agents work through each task while another agent reviews their work in a two-stage process checking spec compliance and code quality. The framework enforces true red/green TDD, YAGNI, and DRY principles. It deletes code written before tests. It creates isolated git worktrees for parallel development. It runs code reviews between tasks and blocks progress on critical issues. Superpowers includes a skills library covering testing, debugging, collaboration, and meta-skills. The brainstorming skill uses Socratic questioning to refine rough ideas. The systematic-debugging skill follows a 4-phase root cause process. The writing-skills skill teaches agents how to create new skills. Available on the official Claude plugin marketplace, Codex CLI, Cursor, OpenCode, GitHub Copilot CLI, and Gemini CLI. The framework is open source under MIT license, with 173k GitHub stars and an active community of contributors and Discord members. Built by Jesse Vincent and the team at Prime Radiant, Superpowers represents a shift from prompt-and-hope to process-driven AI development. It makes coding agents follow the same disciplined methodology that senior engineers use, but faster and more consistently.