Back to list

Superpowers

A working method for coding agents — brainstorm, plan, build with tests, review — on 16 harnesses.

Leon Ford
Author
@Leon Ford
Project ShowcaseDeveloper ToolsOnlineclaude-code
0
0

Project intro

Superpowers is a complete software development methodology for coding agents, built by Jesse Vincent and Prime Radiant. It is a set of composable skills plus bootstrap instructions that make sure the agent actually uses them — instead of jumping straight into code, the agent first asks what you are really trying to build.

The workflow runs in order: brainstorming draws the idea out through questions and plays it back for correction; writing-plans breaks the approved design into bite-sized tasks of two to five minutes, each with exact file paths and verification steps; subagent-driven development dispatches a fresh subagent per task and reviews each one; test-driven development enforces red-green-refactor; and requesting-code-review checks the work before the branch is finished and merged.

More than 319,000 developers starred it, with 28,400 forks, and version 7.0.2 is the latest release. It installs on sixteen harnesses — Claude Code through the official plugin marketplace, plus Codex, Cursor, Gemini CLI, GitHub Copilot CLI, OpenCode, Kimi, Qwen, and more — and ships fifteen skills covering planning, debugging, worktrees, and parallel agents. It is MIT licensed.

It is a method, not a model, so it only works as well as the agent running it. Skills trigger automatically, but a very long session that compacts can lose the bootstrap, and the docs say to start fresh when that happens. Optional telemetry reports only the Superpowers version through a logo image, and it can be switched off with an environment variable.

Actions

Visit project URL

Tech stack

Project info

Project category
Developer Tools
Project status
Online
Collaboration intent
None

Comments (0)

ME
0/1000
Loading comments...