# Superpowers (slim)

> Superpowers is an MIT-licensed library of 15 agent skills and a session-start bootstrap that steer coding agents through brainstorming, a design, a plan, test-driven implementation by subagents and code review. Its README gives install steps for 16 coding agents.

- Full: https://www.anchorterminal.com/tools/superpowers.md (~8,250 tokens) · this version ~1,930 tokens · JSON https://www.anchorterminal.com/tools/superpowers.json · canonical https://www.anchorterminal.com/tools/superpowers
- Index: https://www.anchorterminal.com/llms.txt · API: https://www.anchorterminal.com/api/v1/index.json · MCP: https://www.anchorterminal.com/mcp · Updated: 2026-10-30 · pricing https://www.anchorterminal.com/tools/superpowers#pricing · reviews https://www.anchorterminal.com/tools/superpowers#reviews
- Licence: CC BY 4.0, credit Anchor Terminal and link this page (https://www.anchorterminal.com/terms/#data)

**B · 64.5/100 · rank #408 of 1239 · #7 in Code & developer platforms · not agent-ready · confidence medium**


Assessment: Fifteen MIT-licensed skills that make a coding agent design, plan, test first and review before it finishes, with five releases in 90 days and dated release notes. It has no CI workflow or security policy, its bootstrap insists a skill be invoked before any reply, and its optional visual companion loads a logo from Prime Radiant's site unless disabled.

## Facts

- Kind: Agent framework · vendor: Prime Radiant · category: Code & developer platforms · legal entity: Prime Radiant (no legal form or registration found on its pages); the MIT LICENSE names Jesse Vincent · provenance 58/100
- Auth: None · pricing: Free · x402: no · licence: MIT (Copyright (c) 2025 Jesse Vincent)
- Probe metrics: not measured yet (probes haven't run)
- Surface listed: The obra/superpowers repository at v7.0.1 (main, tree 2ce8aa1), read from raw.githubusercontent.com, GitHub's release, tag, issue and security pages, primeradiant.com, blog.primeradiant.com, blog.fsck.com and claude.com/plugins/superpowers on 30 October 2026. Nothing was installed or run, and no session was opened with it
- How an agent reaches it: As a plugin inside the coding agent, not as an endpoint. The skills sit in the agent's skill list and load on demand, and a SessionStart hook (or a plugin or extension on OpenCode, Pi and Hermes) injects the using-superpowers bootstrap. An outside agent can't call it over a network
- Install routes: Claude Code (official marketplace or obra/superpowers-marketplace), Antigravity, Codex app and CLI (official Codex marketplace), Cursor, Devin CLI, Factory Droid, Gemini CLI, GitHub Copilot CLI, Grok Build CLI, Kimi Code, OpenCode (git+https spec, pin a tag with #v7.0.1), Pi, Qwen Code, Hermes Agent and Muse. No install script; each harness's own plugin manager fetches the repository
- Skills: 15 skills. brainstorming, writing-plans, subagent-driven-development, executing-plans, test-driven-development, requesting-code-review, receiving-code-review, using-git-worktrees, finishing-a-development-branch, dispatching-parallel-agents, systematic-debugging, verification-before-completion, diagnosing-superpowers, writing-skills and using-superpowers. Each description starts with "Use when", about 2,500 characters in all; a test fails any SKILL.md of 20,000 bytes or more
- What runs on the machine: The SessionStart hook is a bash script that reads one SKILL.md and prints JSON, with no network access. On Windows, run-hook.cmd finds Git Bash and, since v7.0.0, ignores bash or where in the current directory. Helper scripts for plans and reviews are bash. The visual companion is a dependency-free Node server bound to 127.0.0.1 by default, with a 4-hour idle timeout
- Data leaving the machine: The visual companion page requests https://primeradiant.com/brand/superpowers-visual-brainstorming-logo.png?v=<version> with no referrer, which the README calls telemetry and says carries no project, prompt or agent details. diagnosing-superpowers files a GitHub issue or builds a scrubbed bundle only after the person approves the exact text. Prompts and code go to the coding agent's model provider as usual
- Work products: Designs and plans under docs/superpowers/, worktrees in .worktrees/ unless told otherwise, and task briefs, implementer reports, review diffs and a progress ledger in a self-ignoring .superpowers/sdd/ directory in the working tree
- Contributions: AGENTS.md says new skills are generally not accepted, PRs must target dev, every PR and issue must disclose the model, harness and plugins used, and agent-written PRs fixing someone else's issue are closed. It states a 94 per cent PR rejection rate, which is the project's own figure
- Company: The repository is under Jesse Vincent's GitHub account. The README says Superpowers is built by him and the rest of Prime Radiant, which its about page calls an AI research lab he founded
- Category: Listed in code, the closest category, though its capabilities are repository and documentation tools. Superpowers is a method a coding agent follows. It works through git branches and worktrees and dispatches subagents, hence code.git and agent.multi-agent

- Scores: Reliability 68, Performance pending, Schema & documentation 71, Agent ergonomics 71, Security & auth 58, Payments & pricing 60, Task success pending, Maintenance & community 83, Transparency & trust 68 · negative events -3 · total over the 7 assessed categories
- Why: Reliability, Local-package reading, as for ai-job-search and the other skill workflows. · Schema & documentation, Framework reading, docs a model can follow. · Agent ergonomics, Framework reading, how much a tool-calling agent has to carry. · Security & auth, Harness reading of the framework checklist, the one harness2 and agent-zero used. · Payments & pricing, Self-hosted rule. · Maintenance & community, v7.0.1 on 21 October 2026, nine days before the check (30). · Transparency & trust, MIT, an OSI licence (30).
- Sources: 26, open questions: 10, both in the full twin
- Capabilities: code.git, agent.multi-agent
- JSON: https://www.anchorterminal.com/api/v1/tools/superpowers.json
- Verify (for the vendor): the badge `https://www.anchorterminal.com/badges/superpowers.svg` or a link to https://www.anchorterminal.com/tools/superpowers from a page on primeradiant.com or one of its subdomains, or the README of github.com/obra/superpowers, then `POST https://www.anchorterminal.com/api/v1/verify` `{"slug", "url"}` or `verify_listing` at /mcp; re-checked weekly, no effect on the grade. Snippets in the full twin.
- Support: Discord https://discord.gg/35wsABTejz; https://github.com/obra/superpowers/issues; more in the full twin

## Before you call it

1. Install it once per coding agent; each harness has its own install command, and the README says a second harness needs a second install.
2. Expect brainstorming to start before any code on a build request. Say so plainly if your human partner wants a quick change done without a design document.
3. Name a model on every subagent dispatch, as the skill asks; an unnamed subagent inherits the session's most expensive model.
4. Set SUPERPOWERS_DISABLE_TELEMETRY=1 before using the visual companion if no request may leave the machine.
5. Keep `.superpowers/sdd/` out of `git clean -fdx`; it holds the progress ledger a resumed run reads.

## Connect

```bash
/plugin install superpowers@claude-plugins-official
```

## Similar tools

| Tool | Grade | Score | Shared capabilities | Slim |
| --- | --- | --- | --- | --- |
| Parallax | E | 38.1 | agent.multi-agent, code.git | https://www.anchorterminal.com/tools/parallax-agents.min.md |
| Unstoppable | F | 34.7 | agent.multi-agent, code.git | https://www.anchorterminal.com/tools/unstoppable.min.md |
| OpenAI Agents SDK | AA | 86.2 | agent.multi-agent | https://www.anchorterminal.com/tools/openai-agents-sdk.min.md |
| Microsoft Agent Framework | A | 82.4 | agent.multi-agent | https://www.anchorterminal.com/tools/microsoft-agent-framework.min.md |
| Pydantic AI | A | 81.9 | agent.multi-agent | https://www.anchorterminal.com/tools/pydantic-ai.min.md |


## Panel reviews (0, desk reviews from public material, no calls made)