
Claude Superpowers: Skills, Plugin Setup, and Limits
By Alex Morgan
MyClaw Editorial
MyClaw
Run Best-in-Class AI Agents Now
Run OpenClaw or Hermes Agent on managed MyClaw hosting, so your AI agent stays online, updated, and ready for real work.
AI Takeaway
- What is Claude Superpowers? Superpowers is a third-party skills library, shipped as the
superpowersplugin. As of 13 November 2026, v7.0.2 is inobra/superpowers, by Jesse Vincent and Prime Radiant, under MIT. - Is it an Anthropic feature? No. It is not a built-in Anthropic feature. It is listed in Anthropic's official Claude Code marketplace, but the source is still
obra/superpowers. - How do you install the plugin in Claude Code? The official README's first route is
/plugin install superpowers@claude-plugins-official. A second route usessuperpowers@superpowers-marketplace. In a session, the command opens details first so you can choose a scope. - What do the Superpowers skills do? Official README skills trigger automatically and follow a mandatory workflow: brainstorm, write a plan, implement with TDD, review, and finish the branch.
- How do you keep the plan after Claude Code closes? Use MyClaw to capture the design, task list, and follow-ups. This article does not claim the plugin runs there, and the homepage is not the installer.
What Claude Superpowers Is — and What It Is Not
Superpowers is a third-party skills library and a software-development methodology for coding agents, shipped as the superpowers plugin. It is not a built-in Anthropic Claude or Claude Code feature.
As of 13 November 2026, the official release is v7.0.2, published 4 November 2026 at 06:17 Asia/Shanghai. The canonical repository is obra/superpowers. Jesse Vincent and Prime Radiant maintain it under the MIT license.
It also appears in Anthropic's official Claude Code marketplace as superpowers@claude-plugins-official, in the author's Superpowers marketplace, and on Claude Marketplace. Official Anthropic plugin docs say the official marketplace includes plugins from partners and other authors, not only plugins Anthropic maintains. A listing, or an "Anthropic verified" label on Claude Marketplace, does not mean Anthropic wrote or controls the plugin's contents.
This is not the archived obra/superpowers-skills repository, and it is not a sibling plugin such as superpowers-chrome or superpowers-lab. It is also not the superpowers-dev branch — the author's marketplace warns you to uninstall other Superpowers versions before that one — and not the VS Code or JetBrains editor plugin. Claude Marketplace, the website, is not the same thing as a catalog you add with /plugin marketplace add.
The official README also lists install paths for other coding agents, but this article stays on Claude Code. If you still need to install Claude Code itself, start with Claude Code for Beginners.
How Superpowers Skills and the Workflow Work
This snapshot follows the official Superpowers README as of 13 November 2026. The v7.0.2 tag and main were identical that day.
In Claude Code, a skill is a SKILL.md file the agent loads when it is relevant, and that you can also run as a command. Superpowers adds a library of these skills, plus a SessionStart hook that bootstraps the using-superpowers instructions. The official README says the agent checks for relevant skills before any task; these are mandatory workflows, not suggestions.
| Step | Skill | What the official README says it does |
|---|---|---|
| 1 | brainstorming | Before building, draw out what you want and why; confirm or write a design |
| 2 | using-git-worktrees | After design approval, isolate work on a new branch and check a clean test baseline |
| 3 | writing-plans | Break the work into short tasks with paths, code, and verification |
| 4 | subagent-driven-development or executing-plans | Per-task subagents with review, or inline execution with one final review |
| 5 | test-driven-development | Red-green-refactor; deletes code written before tests |
| 6 | requesting-code-review | Review against the plan; critical issues block progress |
| 7 | finishing-a-development-branch | Verify tests, then merge, open a PR, keep, or discard, and clean up the worktree |
The official repository lists 15 skill folders. The README groups them as testing (test-driven-development); debugging (systematic-debugging, verification-before-completion, diagnosing-superpowers); collaboration (brainstorming, writing-plans, executing-plans, dispatching-parallel-agents, requesting-code-review, receiving-code-review, using-git-worktrees, finishing-a-development-branch, subagent-driven-development); and meta (writing-skills, using-superpowers). If a session misfires, the official troubleshooting path is diagnosing-superpowers.
The README talks in skill names. The official Claude Marketplace listing, checked on 13 November 2026, also shows slash commands such as /brainstorming and /execute-plan. One README behavior is easy to miss: the test-driven-development skill deletes code written before tests.
How to Install the Claude Superpowers Plugin in Claude Code
You need Claude Code already installed. An interactive session helps, because Anthropic's plugin docs say the official marketplace is added the first time you start an interactive terminal session, unless a policy blocks it.
The official Superpowers README gives two alternative Claude Code routes.
The first installs from Anthropic's official Claude Code marketplace:
/plugin install superpowers@claude-plugins-official
The second adds the author's Superpowers marketplace, then installs from that catalog:
/plugin marketplace add obra/superpowers-marketplace
/plugin install superpowers@superpowers-marketplace
Those slash commands are from the official Superpowers README. The first identifier is Anthropic's official Claude Code marketplace name. Anthropic's Claude Code plugin install docs also document a shell form of the official-marketplace button:
claude plugin install superpowers@claude-plugins-official
In a session, Anthropic's install docs say /plugin install does not install immediately. It opens the plugin details so you can review what will be added — skills, hooks, and any other components — and then choose a scope: user (every project on this machine), project (everyone working in the repository; each person still installs it locally), or local (this repository only). After you confirm, the plugin may be active immediately, or it may need /reload-plugins. Check the Installed tab in /plugin, or run claude plugin list in a shell.
Treat the two routes as alternatives, and confirm in /plugin that only one Superpowers install is enabled. Official plugin docs say an enabled plugin adds skill and command names to context on every turn. The official README does not say you must never install from both marketplaces.
Plugin setup stays in official Claude Code.
Maintenance, Security, and Limits
The official README says updates are "somewhat coding-agent dependent, but are often automatic." Anthropic's plugin docs add that if auto-update is on, files you already reviewed can change later.
Official Anthropic plugin security docs say a plugin can run code with your user privileges, and hooks run as shell commands outside the sandbox. Superpowers ships a SessionStart hook on startup, clear, and compact. Before you confirm install, review the details pane and the plugin source, including hooks/hooks.json. The catalogs sit in different trust tiers: claude-plugins-official is official-tier, and superpowers-marketplace is third-party.
While the plugin is enabled, official docs say skill, agent, and command names and descriptions sit in context on every turn, even when unused, and count toward usage. Official-marketplace details can show a context-cost estimate.
The official README describes optional telemetry: brainstorming's optional visual companion loads a Prime Radiant logo that includes the Superpowers version, not project or prompt details. Disable it with SUPERPOWERS_DISABLE_TELEMETRY, or with Claude Code's DISABLE_TELEMETRY and CLAUDE_CODE_DISABLE_NONESSENTIAL_TRAFFIC.
Test-driven development may delete pre-test code, and worktrees and subagents change the workspace. If a session goes wrong, ask the agent to figure out what went wrong with Superpowers in this session, which should invoke diagnosing-superpowers.
The license remains MIT. The README offers optional commercial support, but no public price was listed as of 13 November 2026.
Plan the Surrounding Work in MyClaw

A Superpowers session in Claude Code can leave you with an agreed design, a task list, and follow-up checks. That work is easier to keep if it does not depend on the laptop session staying open, so the next coding session can start from the same plan. MyClaw is a place to plan and run that surrounding workflow. This article does not claim the Superpowers plugin is installed, supported, or unavailable there. Keep plugin setup in official Claude Code; the homepage is not the install path.
- After the Claude Code session has an agreed design or task list, capture the follow-up work, constraints, and checks in MyClaw.
- Turn the parts that should keep running after you close the laptop into a hosted agent workflow — reminders, status checks, or messages.
- Run that plan in MyClaw, and use official Claude Code with the Superpowers plugin when the job needs this exact skills workflow.
Conclusion
As of 13 November 2026, Claude Superpowers is the v7.0.2 release of a third-party plugin in obra/superpowers. It is listed in Anthropic's official Claude Code marketplace, but it is not a built-in Anthropic feature, and a listing does not mean Anthropic maintains the code.
If you want this skills workflow, install the plugin in official Claude Code, review the details and scope first, and keep the official marketplace route distinct from the author's marketplace. Use MyClaw to plan and run the work around that session.
Skip the Setup, Run Best-in-Class AI Agents Now
Launch a managed OpenClaw or Hermes Agent workspace in minutes, with always-on hosting, updates, and support handled by MyClaw.