Huashu Design
alchaincyf · Skill
Turns a one-line request in your AI agent into a finished design: a launch animation, a clickable app prototype, an editable slide deck, an infographic or a se…
New29.4k
Overview
Jesse Vincent’s skills that give your coding agent a working method: it asks what you want, gets a design approved, writes a step-by-step plan, then builds it test first with code review.
Superpowers is an open-source project on GitHub in the Developer category, listed by Mindket. It is free. This is version 2, last updated on . Install it from GitHub as its README says. Get Mindket’s setup notes on your Mindket account, as a zip.
At a glance
- Runs in
- Claude Code: from Anthropic’s official marketplace or from the maker’s own
- Cursor: one line in the Agent chat
- GitHub Copilot CLI: through the maker’s marketplace
- Codex, Gemini CLI and OpenCode: each has its own steps in the kit’s notes
- Start with
/plugin install superpowers@claude-plugins-official
Platform
Claude Code
Cursor
GitHub Copilot
Codex
Gemini CLI
OpenCode
And 1 more in Where to use it
About
What it is
Superpowers is a set of skills from Jesse Vincent and the team at Prime Radiant that gives a coding agent a full method for building software. Before any code, the agent asks what you are trying to do and gets a design approved. It then writes a plan of small tasks and builds them test first, with review between tasks. The skills start on their own, since the agent checks for a matching one before each task.
- You describe what you want to build
- Questions, then a design doc you approve
- A plan of tasks that take 2 to 5 minutes each
- Subagents build each task test first, with review
- Tests verified, then you pick merge, PR, keep or discard
Who it suits
Developers who use coding agents for real features and want them to plan and test before they write code. Teams that already practise test-driven development and code review will recognize the method.
| Industry | Who uses it | How they use it | Example |
|---|---|---|---|
| Software and SaaS companies | Product developers | Ask for a feature, approve the design, then let the agent build it test first | A developer new to a codebase asks for a feature and learns the code through the agent’s questions and design doc. |
| Teams with code review rules | Tech leads | Keep the committed design docs and the review between tasks as the team’s record | A team reads the design docs in its repository to see why each feature was built the way it was. |
| Startups | Solo founders | Hold the agent to a plan, so it does not rewrite large parts of an app | A founder asks for a small change, and the plan keeps it small. |
| Teams that run agents for hours | Developers who hand long work to subagents | Let a fresh subagent take each task, with a review after every one | A developer leaves subagents on a long plan and reviews each task as it lands. |
If you mostly want quick one-off scripts, the questions and plans will feel slow.
Where to use it
The README gives install steps for 16 coding agents and apps, and each one needs its own install.
- Claude Code: from Anthropic’s official marketplace or from the maker’s own
- Cursor: one line in the Agent chat
- GitHub Copilot CLI: through the maker’s marketplace
- Codex, Gemini CLI and OpenCode: each has its own steps in the kit’s notes
- More agents: Antigravity, Devin CLI, Factory Droid, Grok Build CLI, Kimi Code, Pi, Qwen Code, Hermes Agent and Muse
It fits work where a wrong start costs more than a few questions:
- A feature that needs a design: the agent asks first and waits for your approval
- A long plan: subagents carry it task by task while you review their work
- A stubborn bug: systematic-debugging works toward the root cause in four phases
- A session that went off plan: diagnosing-superpowers reads the transcript and reports with line-level evidence
How to set it up
Get it saves a zip of Mindket’s setup notes in your downloads folder, with each agent’s install steps from obra’s README.
| Need | Detail |
|---|---|
| A coding agent | One the README covers. Each agent you use needs its own install |
| A plugin source for Claude Code | Anthropic’s official marketplace, or obra’s own marketplace with some related plugins |
| A git repository | brainstorming commits its design doc there, and using-git-worktrees works on a new branch |
- Run the install line for your agent. Each agent you use needs its own install.
- Restart open sessions where the README says to, as on Hermes Agent and Muse.
- In a fresh session, ask for something to build. On a working install, brainstorming starts before any code is written.
/plugin install superpowers@claude-plugins-official
The maker’s own marketplace carries Superpowers too, with some related plugins.
/plugin marketplace add obra/superpowers-marketplace /plugin install superpowers@superpowers-marketplace
In Cursor’s Agent chat it is one line, and GitHub Copilot CLI takes the same marketplace. The kit’s notes carry the lines for the other agents.
/add-plugin superpowers
copilot plugin marketplace add obra/superpowers-marketplace copilot plugin install superpowers@superpowers-marketplace
How to use it
Describe what you want, and the skills take over in this order.
| Step | Skill | What it does |
|---|---|---|
| 1 | brainstorming | Draws out what you want and why, then plays it back for correction |
| 2 | using-git-worktrees | Sets up an isolated workspace on a new branch and checks for a clean test baseline |
| 3 | writing-plans | Breaks the work into small tasks, each with exact file paths and checks |
| 4 | subagent-driven-development | Sends a fresh subagent per task with a review after each, or executing-plans builds inline with one review at the end |
| 5 | test-driven-development | Writes a failing test first, watches it fail, then writes the least code that passes |
| 6 | requesting-code-review | Checks work against the plan between tasks, and critical issues block progress |
| 7 | finishing-a-development-branch | Verifies the tests and offers merge, PR, keep or discard |
Bugs go through systematic-debugging, a four-phase root cause process. When a session misbehaves, ask the agent the line below and the diagnosing-superpowers skill reads the transcript.
figure out what went wrong with superpowers in this session
Example
In a project folder, you open your coding agent and ask for a small app. The README uses the same request to check an install.
Let's make a react todo list
No code at first. The agent asks what you want and why, lays out two or three ways to build it and has you pick one. A short design doc follows for your approval, committed to docs/superpowers/specs in a software repository. Then comes a plan of small tasks with exact file paths, built test first.
For newcomers and experts
If you are new. Left alone, a coding agent often starts typing code at once. Superpowers makes it work like a careful engineer: ask, agree on a design, plan, then build with tests. Try it on a small feature in a test project and answer its questions as they come.
For experts. Each step is its own Markdown skill, so you can read or edit one without touching the rest. Plans are written so a builder with no project context could follow them, and they lean on red-green TDD, YAGNI and DRY. The README says an agent often works on its own for a couple of hours without leaving the plan. The method is strict by design. Its price is extra turns and tokens on changes too small to need a design doc.
Good to know
The skills change how your agent works, and two of them change your files.
- What runs: the skills are Markdown files that run inside your agent
- Session start: on Claude Code, a SessionStart hook loads the using-superpowers skill when a session starts, is cleared or is compacted
- Code it deletes: the test-driven-development skill deletes code written before its test
- Your repository: brainstorming commits its design doc there
- Visual companion: an optional local server that loads a Prime Radiant logo, a request that carries your Superpowers version and nothing about your project
- Turning that off: SUPERPOWERS_DISABLE_TELEMETRY, or Claude Code’s DISABLE_TELEMETRY and CLAUDE_CODE_DISABLE_NONESSENTIAL_TRAFFIC opt-outs
- Updates: they depend on the agent and are often automatic, and the README gives no uninstall steps
- License: MIT
README and files
# Superpowers setup notes This kit holds setup notes written by Mindket, not the skills themselves. The Superpowers skills and plugin install from obra's repository: https://github.com/obra/superpowers License: MIT, Copyright (c) 2025 Jesse Vincent. Source: https://github.com/obra/superpowers/blob/main/LICENSE Made by Jesse Vincent and the team at Prime Radiant. Listed by Mindket. ## What installs Superpowers is a library of skills for coding agents, plus a startup step that tells the agent to use them. Skill list from the README section "What's Inside" (https://github.com/obra/superpowers#whats-inside): - 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 - Meta: writing-skills, using-superpowers On Claude Code, the plugin adds a SessionStart hook that runs when a session starts, is cleared or is compacted. It reads the using-superpowers skill and passes it to the session as context. Source: https://github.com/obra/superpowers/blob/main/hooks/hooks.json ## Pick your agent The README says installation differs by agent. If you use more than one, install Superpowers separately in each. Source: https://github.com/obra/superpowers#installation Each file below has the install steps the maker gives for that agent: - setup/claude-code.md - setup/codex.md - setup/cursor.md - setup/gemini-cli.md - setup/github-copilot-cli.md - setup/opencode.md - setup/other-agents.md: Antigravity, Devin CLI, Factory Droid, Grok Build CLI, Kimi Code, Pi, Qwen Code, Hermes Agent and Muse ## Updating The README says updates depend on the coding agent and are often automatic. Where the maker gives an update command, that agent's file lists it. Source: https://github.com/obra/superpowers#updating ## Removing The maker's README gives no uninstall steps. ## Telemetry Brainstorming has an optional visual companion. By default it loads the Prime Radiant logo from the Prime Radiant website, and that request includes your Superpowers version. The README says it carries no details about your project, prompt or coding agent. To turn it off, set SUPERPOWERS_DISABLE_TELEMETRY to any true value. Claude Code's DISABLE_TELEMETRY and CLAUDE_CODE_DISABLE_NONESSENTIAL_TRAFFIC opt-outs also work. Source: https://github.com/obra/superpowers#visual-companion-telemetry ## When something goes wrong Ask your coding agent to "figure out what went wrong with superpowers in this session". That starts the diagnosing-superpowers skill. It reads the session transcript, reports what happened with evidence and, if you want, packages a scrubbed bundle for a bug report. To check an earlier session, name it: "figure out what went wrong with superpowers in session `<id>`". Bug reports go to https://github.com/obra/superpowers/issues. Source: https://github.com/obra/superpowers#when-something-goes-wrong
README.md 3 KB setup/claude-code.md 1 KB setup/codex.md 670 B setup/cursor.md 381 B setup/gemini-cli.md 309 B setup/github-copilot-cli.md 451 B setup/opencode.md 2 KB setup/other-agents.md 5 KB
What’s new
Updated by Mindket to the current version of this kit.
Reviews
What people say
Posts and comments that name this project on Hacker News, Bluesky, X, Reddit, Facebook and YouTube, found by a weekly search and on Trendshift. Mindket doesn’t write or check them. Open one to read it where it was posted.
anaq42Show HN: JDS – a Copilot skill suite for structuring AI coding behaviorA few months ago I stumbled on obra's superpowers repository https://github.com/obra/superpowers. I really liked the approach and idea that you enforce discipline for your agent through a skill-based workflow. Even though coding agents (copilot included) have become a lot better at natively…9 pointsLeo Lapworthgithub.com/obra/superpowers and github.com/ovid/paad together work well for me flow is sp:brainstorm paad:pushback sp:write-plan paad:alignment sp:subagent-implementation - both have other useful skills in them but that is main flow10 likes · 2 replies@zyonbuilds@pingdotgg @mattpocockuk 4. Claude Code Game Studios turns Claude Code into a whole game dev studio. → 49 agents in a studio-style hierarchy → 72 workflow skills +5.3k stars this week, 36.7k total. MIT. https://t.co/fjqIosUsg5dsifryShow HN: 127 PRs to Prod this wknd with 18 AI agents: metaswarm. MIT licensedA few weeks ago I posted about GoodToGo https://news.ycombinator.com/item?id=46656759 - a tool that gives AI agents a deterministic answer to "is this PR ready to merge?" Several people asked about the larger orchestration system I mentioned. This is that system. I got tired of being a project…5 points · 2 repliesEdy | Indie BuilderToday's GitHub trending: everyone is building skill packs for AI agents. - mattpocock/skills: skills from Matt Pocock's own .agents folder (+883 today) - obra/superpowers: agent skills framework plus a dev method (+665) - Orkas-AI/Orkas: local-first desktop AI team (+1,044) #buildinpublic #AI7 likes · 1 reply@wushi123321Your coding agent jumps into code before it knows what you want. Superpowers is an open methodology plugin: brainstorm → plan → subagent tasks → TDD, for Claude Code, Cursor, Codex, Gemini, and more. 313.8k stars. https://t.co/ghL5s1qLY4 #GitHub #OpenSource https://t.co/9xiYcNwUDbiamandoniShow HN: Arxitect – Agentic Plugin for Architecture and Design PatternsModern coding agents are getting exceptionally good at implementing a given coding task. And with validation-in-the-loop, you can be reasonably confident they will implement a correct solution. However, their implementation often leaves a lot to be desired. It doesn't adhere to the decades of…3 points@fenju.bsky.socialToday's GitHub Trending signals a shift: agents must go from "can it work?" to "can it work well, with standards?" pbakaus/impeccable & obra/superpowers prove devs want engineering-grade specs, not improvisation. Same for multimodal: speech, OCR, translation, proofreading shouldn't be prompt-enginee3 likes · 1 reply@xqliu你是不是也遇到过：跟 AI 说一句「帮我加个功能」，它一个问题都不问就开写，写完才发现根本不是你要的。 obra/superpowers 专治这个。它是一套给编程 agent 用的 skill，外加一套开发方法。这周又涨了四千多星，总星数过了三十一万。 装上以后，agent 干活的顺序变了： 先不碰代码，反过来问你到底想做什么。问清需求后，分成一小段一小段给你看，每段都短到你真能读完。 你点头了，它才写实现计划。作者对这份计划的要求是：一个热情但没品味、没判断力、不了解项目、还讨厌写测试的新人，拿着它也能照做。 你说 go，它再派子 agent 一个任务一个任务往下做，每个子 agent…iamandoniShow HN: Arxitect – Claude Code plugin for software design principlesShow HN: Arxitect – Claude Code plugin for software design principles Modern coding agents are getting exceptionally good at implementing a given coding task. And with validation-in-the-loop, you can be reasonably confident they will implement a correct solution. However, their implementation…2 points@antspaceai.bsky.socialobra/superpowers (Plugin, 313,095 GitHub stars): An agentic skills framework and methodology for software development Install in Claude Code: /plugin marketplace add obra/superpowers https://antspace.ai/go/i4r99jpi1 like@AmanSha017493355 free GitHub AI repos that feel illegal to know about 👇 Full breakdown in the video. All open source on @GitHub. https://t.co/uwxjKw1ewc
More from Mindket
See all from MindketGuizang PPT Skill
op7418 · Skill
Skill that makes single-file HTML slide decks, slide illustrations and platform covers in two visual styles, with a presenter mode for rehearsal. For speakers…
New29.8k
Claude Code Skills & Plugins
alirezarezvani · Skill
Adds 411 ready-made skills for Claude Code and other coding agents, grouped by role from engineering and product to marketing, finance and compliance. Installs…
New29.8k
Similar projects
See all in DeveloperSupabase Agent Skills
supabase · Skill
Supabase’s own agent skills: one for any Supabase task, one for Postgres best practices from queries and indexes to connections and row level security. They in…
New3023
buzz
block · Kit
Self-hostable workspace where people and AI agents share the same rooms, built on a Nostr relay that logs every action as a signed event. For teams that want t…
New37.3k
plannotator
backnotprop · Kit
Opens your AI coding agent's plans, specs and code changes in a browser so you can annotate or comment on them and send the feedback straight back to the agent.
New10.5k