PluginRising

Superpowers

Plugin : A set of capabilities installed into an AI application or a host · requires claude-code

Complete development methodology for agents, built on composable skills that activate automatically: brainstorming, TDD, debugging, review.

The most popular skills plugin; it changes the way the agent works, not what it can access.

GitHub · 304,688 ★

Overall score

8.3 /10

Security: one caveat, see the card

Utility

9.0

Reliability

10.0

Security

One caveat

Ease

1 click · Comfortable

Security reflects where data goes, requested permissions and incidents on record at the verification date. It is not a guarantee of zero risk: check the deployment mode and the connections you enable.

What it does, concretely

Coding agents like Claude Code tend to jump straight into writing code without fully understanding the requirement, leading to costly back-and-forth and results that drift from what was intended. Superpowers enforces a structured methodology from the very start, before the first line of code is written.

The plugin installs a library of composable skills that activate automatically based on context: Socratic brainstorming to refine the requirement, task-by-task implementation planning, test-driven development (red-green-refactor), code review between each task, and Git branch management. The agent can then work autonomously for hours without deviating from the agreed plan.

Superpowers is aimed at developers who use coding agents daily and want to govern their behaviour without repeating instructions every session. It is compatible with many environments: Claude Code, Codex, Cursor, Gemini CLI, GitHub Copilot CLI, Devin, and others.

It offers little to those looking to connect the agent to external data sources or third-party tools: Superpowers acts on how the agent reasons and works, not on what it can access.

What it offers

·Socratic brainstorming: the agent asks questions to refine the requirement before writing anything, then presents the design in readable sections for validation.
·Detailed plan writing: each task takes 2 to 5 minutes and includes exact file paths, complete code, and verification steps.
·Test-driven development (TDD): the red-green-refactor cycle is mandatory; code written before tests is deleted.
·Subagent-driven development: a separate subagent is launched for each task, with a two-stage review (spec compliance, then code quality).
·Automatic code review between tasks: critical issues block progress.
·Systematic four-phase debugging to identify the root cause before fixing.
·Git branch management with worktrees: isolated workspace on a new branch, clean test baseline verified before starting.
·Creation of custom skills via the writing-skills skill, following the project's best practices.

Our tips

If you use Superpowers in multiple environments (Claude Code and Codex for example), install it separately in each one: installations are not shared.
To disable the visual telemetry linked to the Prime Radiant logo, set the environment variable SUPERPOWERS_DISABLE_TELEMETRY to any truthy value; the plugin also honours Claude Code's DISABLE_TELEMETRY and CLAUDE_CODE_DISABLE_NONESSENTIAL_TRAFFIC variables.
Skills activate automatically: there is no need to call them manually. Let the agent take the initiative and step in during the design and plan validation stages.
On Hermes Agent, a very long session that compacts beyond the first turn loses the bootstrap: start a fresh session if skills stop triggering.
Join the community Discord for support and release announcements, as updates may be automatic depending on the environment used.

Watch out

On Hermes Agent, a very long session that compacts beyond the first exchange loses the Superpowers bootstrap, disabling the skills: a fresh session must be started.
Superpowers deeply modifies agent behaviour through instructions it injects: it is advisable to read the plugin content before installing it, as with any skill that drives the agent.

In practice

Price: Free
Commercial use: Commercial use: yes
Available on: claude-code, codex, opencode
Host: claude-code
Licence: MIT

Security and data

Rating: Security: one caveat, see the card
Your data: Data stays local
Maintenance: Active
Last release: 21 October 2026

Links and repository checked on 25 October 2026 · added on 15 July 2025

How to install Superpowers

1. Check which environment you use (Claude Code, Codex, Cursor, etc.): installation differs by host. 2. For Claude Code, install from the official Anthropic marketplace with the command: /plugin install superpowers@claude-plugins-official 3. For Codex CLI, open the search interface with /plugins, search for "superpowers" and select "Install Plugin". 4. For other environments (Cursor, Gemini CLI, Devin, etc.), refer to the GitHub README for the command specific to your host. 5. If you use multiple environments, install Superpowers separately in each one. 6. Caution: Superpowers drives agent behaviour through injected instructions; read the plugin content before enabling it.

/plugin install superpowers@claude-plugins-official

In the same category

Whole category Frame (methods, guardrails) →