Radar check
What we found when we opened it. Every bar is one fact, with its date.
Last code pushed 23 Oct 2026, 4 days before this check.
Latest release v7.0.1 on 21 Oct 2026.
Public since October 2025, about 1 year.
306,444 stars and 27,278 forks.
6 commits and 251 new issues in the last 90 days.
44 contributors. 78 pull requests merged in 90 days.
No security policy file says how to report a problem. No published security advisories.
How the bars are scored
- Alive. Days since the last code push: a week or less 100, a month 90, three months 70, six months 50, a year 30, eighteen months 15, older 0.
- Releases. Age of the latest GitHub release, on a similar scale; no bar when the project publishes no releases.
- Maturity. Age of the project: eight years 100, five 90, three 70, two 55, one 40, six months 25, younger 10.
- Adoption. Stars on a log scale: 100,000 stars is 100, 10,000 is 80, 1,000 is 60, 100 is 40.
- Activity. Commits in the last 90 days: 300 or more 100, 100 85, 30 65, 10 45, any 25.
- Community. People who have contributed: 100 or more 100, 30 80, 10 60, 5 40, 2 25, one 10.
- Security. A published security policy 80, none 40; published advisories are listed, not scored.
Checked by Repo Radar on 28 Oct 2026 from the repository on GitHub.
In simple English
When you use an AI coding assistant, it often rushes to write messy code without a plan. This tool acts as a set of rules and skills that forces your assistant to slow down. It makes the assistant ask you what you want first, write a clear plan, and test its work properly. You install it into whichever coding assistant app or command-line tool you already use. It costs nothing except for the time it takes to review the agent's plans.
What it is
Superpowers is a collection of prompt-based skills and workflows that guide AI coding assistants through software development. It intercepts the start of a session to enforce brainstorming, implementation planning, and strict testing cycles.
What it is good for
Solo builders who want their AI coding assistants to write better, more thoroughly tested code instead of guessing and making a mess.
Summary (Key Points)from the README
Key features
- Design before code: The agent asks what you're trying to do, then shows the spec in chunks short enough to read and digest before any code is written.
- Bite-sized plans: The writing-plans skill breaks work into tasks of 2-5 minutes each, with exact file paths, complete code and verification steps.
- Strict TDD: test-driven-development enforces RED-GREEN-REFACTOR and deletes code written before its tests.
- Subagent execution: subagent-driven-development dispatches a fresh subagent per task, with two-stage review (spec compliance, then code quality).
- Inline option: executing-plans runs every task in one session with a single final review of the whole branch, which is the cheapest mode.
- Isolated workspaces: using-git-worktrees creates a new branch, runs project setup and verifies a clean test baseline.
- Automatic triggering: Skills fire on their own, so you don't need special commands. The README says the agent can often work autonomously for a couple hours without deviating from the plan.
- Session diagnostics: diagnosing-superpowers reads a session transcript, reports what went wrong with line-level evidence, and can package a scrubbed bundle for a bug report.
Ways to install
| Platform or method | What it needs | Effort |
|---|---|---|
| Claude Code | Plugin commands from the official or Superpowers marketplace | Two commands |
| Antigravity | `agy plugin install` with the repo URL | One command |
| Codex App | Plugins sidebar in the app | Few clicks |
| Codex CLI | `/plugins` search inside the CLI | Few steps |
| Cursor | `/add-plugin superpowers` in Agent chat, or marketplace search | One command or a search |
| Devin CLI | `devin plugins install obra/superpowers` | One command |
| Factory Droid | Add the marketplace, then install the plugin | Two commands |
| Gemini CLI | `gemini extensions install` with the repo URL | One command |
| GitHub Copilot CLI | Add the marketplace, then install the plugin | Two commands |
| Grok Build CLI | `grok plugin install` or `/marketplace` in the TUI | One command or a search |
| Kimi Code | `/plugins` manager, or install from the repo URL | Few steps |
| OpenCode | Tell the agent to fetch and follow an install file | Prompt-based |
| Pi | `pi install git:` with the repo | One command |
| Qwen Code | `qwen extensions install obra/superpowers` | One command |
| Hermes Agent | `hermes plugins install ... --enable`, then restart sessions | One command plus restart |
| Muse | Local checkout or clone, then install and approve the plugin | Several commands |
Important
- Install per harness: Installation differs by harness, so each one you use needs its own install. OpenCode needs one even if you already use Superpowers elsewhere.
- Hermes bootstrap loss: Hermes has no post-compaction hook, so a long session that compacts can lose the bootstrap. Start a fresh session if skills stop triggering.
- Muse approval: Hooks need approval on first install, and sessions must be restarted after installing.
- Contribution limits: New skills are generally not accepted, skill updates must work across all supported agents, and agent-written PRs for someone else's issue are not accepted unless a maintainer asks.
- Telemetry on by default: The brainstorming visual companion loads a Prime Radiant logo from the website, which includes the Superpowers version. Set `SUPERPOWERS_DISABLE_TELEMETRY` to a true value to disable it; Claude Code's opt-out variables are also honored.
- Updates vary: Updates are described as somewhat agent-dependent, though often automatic.
Written by an AI model from the project’s README; not part of the review, and it can be wrong.
- Forces AI assistants to follow proven software engineering steps instead of guessing.
- Works across many popular AI coding tools like Claude Code, Cursor, and Gemini CLI.
- Enforces test-driven development by making the assistant write failing tests first.
- Requires an existing AI coding assistant tool to function at all.
- Can make your AI assistant slower because it insists on brainstorming and planning first.
- Debugging session misbehaviors requires asking the AI to inspect its own transcripts.
Ad-hoc prompting and trial-and-error AI coding sessions
Claude Code
- anthropics/claude-code154k stars
An official command-line coding assistant that can run plugins like Superpowers.
- musistudio/claude-code-router38k stars
One local address that many coding assistants can talk to, pointed at whichever model the operator chooses.
- Kong/insomnia40k stars
A desktop application for sending requests to an interface and reading what comes back.
- microsoft/playwright-mcp39k stars
A server that lets an AI assistant drive a real browser, using the page's own structure rather than screenshots.
/plugin install superpowers@claude-plugins-official