Skip to content
Back to search

superhero-skills

MarketplaceWorks with Claude CodeGitHub

Product Discovery, Product Design and Engineering skills for product builders, from Superhero.tech.

Safety scan

A static text scan, not a guarantee. Select the badge to see what was scanned and every finding.

Checked today
Badge for your README
findsafeskills safety rating

Markdown

[![findsafeskills safety rating](/api/listings/superhero-tech--superhero-skills/badge.svg)](/listings/superhero-tech--superhero-skills)

HTML

<a href="/listings/superhero-tech--superhero-skills"><img src="/api/listings/superhero-tech--superhero-skills/badge.svg" alt="findsafeskills safety rating" /></a>

Install

Install instructions from the README

As written by the repo's author. findsafeskills has not verified these steps.

If you use a desktop app, you can add the marketplace from the interface without cloning this repository or using a terminal.

ChatGPT Desktop

Add the marketplace directly from the app:

1. Open **Plugins** from the left sidebar. 2. Select **Add+** in the top-right corner. 3. Select **Add marketplace**. 4. Paste [https://github.com/superhero-tech/superhero-skills/](https://github.com/superhero-tech/superhero-skills/) and confirm. 5. The **Superhero.tech** marketplace is now available in Plugins. 6. Install **Superhero Skills**.

Claude Desktop

Add the marketplace directly from the app:

1. Select your username in the bottom-right corner, then select **Settings**. 2. Select **Plugins**. 3. Select **Add+** in the top-right corner. 4. Paste [https://github.com/superhero-tech/superhero-skills/](https://github.com/superhero-tech/superhero-skills/) and confirm. 5. The **Superhero.tech** marketplace is now available in Plugins. 6. Install **Superhero Skills**.

Claude Code

Run these commands inside Claude Code:

/plugin marketplace add superhero-tech/superhero-skills
/plugin install product-discovery@superhero-skills

If Claude Code asks you to reload the plugin, run `/reload-plugins` or start a new session.

Invoke a skill explicitly, for example:

/product-discovery:supersentence
/product-discovery:supersprint
/product-discovery:pres-prioritization
/product-discovery:experiment-design
/product-discovery:ab-test-design
/product-discovery:customer-research
/product-discovery:affinity-mapping
/product-discovery:desk-research
/product-discovery:interview-plan
/product-discovery:interview-simulator
/product-discovery:data-analysis
/product-discovery:code-event-storming
/product-discovery:production-readiness
/product-discovery:super-landing-page
/product-discovery:visual-direction-research
/product-discovery:document-design-system-components
/product-discovery:build-figma-library

Codex

Run these commands in your shell:

codex plugin marketplace add superhero-tech/superhero-skills
codex plugin add product-discovery@superhero-skills

Start a new Codex session after installation. Version `1.11.0` adds `visual-direction-research` and `document-design-system-components` alongside the existing `build-figma-library`; check that your installed plugin has updated to this version or later to use the complete design workflow. Version `1.14.0` adds `code-event-storming`; version `1.15.0` adds `production-readiness`. Version `1.16.0` adds `ab-test-design`. You can also open `/plugins` to browse the `Superhero Skills` marketplace.

Invoke a skill explicitly, for example:

$product-discovery:supersentence
$product-discovery:supersprint
$product-discovery:pres-prioritization
$product-discovery:experiment-design
$product-discovery:ab-test-design
$product-discovery:customer-research
$product-discovery:affinity-mapping
$product-discovery:desk-research
$product-discovery:interview-plan
$product-discovery:interview-simulator
$product-discovery:data-analysis
$product-discovery:code-event-storming
$product-discovery:production-readiness
$product-discovery:super-landing-page
$product-discovery:visual-direction-research
$product-discovery:document-design-system-components
$product-discovery:build-figma-library

Install individual skills

To select individual skills and install them in Codex, Claude Code, OpenCode or another supported agent, use [`skills.sh`](https://skills.sh):

npx skills@latest add superhero-tech/superhero-skills

Add `--global` if you want the skills available across all your projects. If you use Claude Code, choose either the plugin or individual installation to avoid installing every skill twice.

What was scanned

The files the safety scan read, as written by the repository's author and shown as plain text. Lines with a finding are marked.

.claude-plugin/marketplace.json · Manifest · 45 lines
  1. {
  2. "name": "superhero-skills",
  3. "owner": {
  4. "name": "Superhero.tech",
  5. "email": "superhero@superhero.tech",
  6. "url": "https://superhero.tech"
  7. },
  8. "description": "Product Discovery, Product Design and Engineering skills for product builders, from Superhero.tech.",
  9. "plugins": [
  10. {
  11. "name": "product-discovery",
  12. "displayName": "Superhero Skills",
  13. "source": "./plugins/product-discovery",
  14. "description": "Product discovery and design workflows: business ideas, Supersprint discovery cycles, customer research, affinity mapping, code-based EventStorming, desk research, interview planning and simulation, data analysis, PRES prioritization, experiment design, A/B test planning with traffic and duration checks, art-directed landing pages, visual direction research, design-system component documentation, and reusable Figma libraries. Production-readiness guides release preparation with an evidence-backed offline HTML checklist.",
  15. "category": "product-management",
  16. "tags": [
  17. "discovery",
  18. "research",
  19. "customer-interviews",
  20. "data-analysis",
  21. "experiments",
  22. "supersprint",
  23. "landing-page",
  24. "supersentence",
  25. "interview-simulator",
  26. "figma",
  27. "design-system",
  28. "pres",
  29. "visual-direction",
  30. "moodboard",
  31. "component-documentation",
  32. "event-storming",
  33. "engineering",
  34. "production-readiness",
  35. "ab-testing"
  36. ],
  37. "version": "1.16.0",
  38. "author": {
  39. "name": "Superhero.tech",
  40. "email": "superhero@superhero.tech",
  41. "url": "https://superhero.tech"
  42. }
  43. }
  44. ]
  45. }
README.md · README · 261 lines
  1. # Superhero Skills
  2. AI made it cheap to generate code, designs, research and copy. It did not make product judgment cheap. Point it at the wrong problem and it will help you ship the wrong thing faster.
  3. Product judgment becomes the bottleneck. Someone still has to decide what is worth building, judge the output and take the work from the first signal to a shipped product. We call that person a Product Builder.
  4. Product Builders are deep in one discipline and capable across two or three adjacent ones. In practice, that can be a PM who prototypes instead of waiting for engineering, an engineer who runs interviews instead of waiting for research, or a designer who ships working code instead of handing off mockups. AI helps them close skill gaps without handing over judgment.
  5. We built these skills from 32+ years of combined experience building global products at Booking.com, CD Projekt, Displate and GOG.com. Versions of these workflows have been used by 1,800 Product Builders in [AI Product Heroes](https://aiproductheroes.pl).
  6. ## The Product Builder OS
  7. Superhero Skills are one half of a larger system.
  8. | Layer | What it does |
  9. |---|---|
  10. | [Super Brain](https://github.com/superhero-tech/super-brain) | Keeps your context, knowledge, projects and decisions, so the agent does not start every conversation from zero. |
  11. | Superhero Skills | Give the agent focused, tested ways to work through product problems with you. |
  12. Together, these two layers make up the Product Builder OS. Superhero Skills also works on its own: Super Brain or a personal knowledge vault is not required.
  13. ## The Super Loop: product development for the AI era
  14. We are building this skill set around a six-step loop:
  15. ```text
  16. Sense -> Frame -> Prototype -> Validate -> Ship -> Learn -> repeat
  17. ```
  18. | Step | What happens |
  19. |---|---|
  20. | Sense | Collect signals from customers, data and the market. |
  21. | Frame | Decide which problem is worth solving. |
  22. | Prototype | Make the idea real enough to learn from. |
  23. | Validate | Put it in front of real users, not synthetic ones. |
  24. | Ship | Release a small, reversible version. |
  25. | Learn | Use evidence to iterate, scale or kill it. |
  26. Traditional teams run this loop in quarters, slowed down by a handoff at every step. Product Builders can run it in days. AI accelerates all six steps, but it does not own Frame, Validate or the decision in Learn.
  27. ## Available skills
  28. **Superhero Skills** is one plugin with three skill folders: **Product Discovery**, **Product Design** and **Engineering**. The technical installation ID and command namespace remain `product-discovery`, so existing installations can update normally. Folder grouping organizes the repository; the host may display a flat list of skills.
  29. The skills need no separate API keys, personal knowledge vault or private development repository. Supply your own project context and evidence. Requirements depend on the work:
  30. - A host that can read skills and save files in your chosen project.
  31. - Python 3 for the bundled tree, affinity-map and recommendation renderers; these use only the standard library. The optional customer-opinion counter also uses standard Python.
  32. - For `ab-test-design`, Python 3 to render a standalone HTML card. Node.js is needed only for numerical checks and the optional command-line summary; the interactive calculator runs in the browser.
  33. - For `code-event-storming`, access to the project's source code and Python 3.9 or later. The renderer uses the standard library, and the generated board works offline in a browser with JavaScript enabled.
  34. - For `production-readiness`, local project access and Python 3.9 or later. The renderer uses only the standard library; the single HTML report works offline. JavaScript enables optional viewing filters.
  35. - For data analysis, pandas, matplotlib and seaborn; DuckDB for SQL, and readers such as openpyxl or pyarrow when the input format needs them. The plugin provides instructions and scripts, not a Python distribution or these libraries.
  36. - Live web search and readable original sources for `desk-research` and `customer-research`.
  37. - For `visual-direction-research`, live web search, browser access to inspect rendered websites, and screenshot capture. Optional PDF export needs a PDF-capable workflow.
  38. - For `document-design-system-components`, visual source materials such as screenshots, mockups, PDF or HTML specifications, and a host that can inspect their images. Figma access is not required.
  39. - Browser support for automatic HTML previews. If preview is blocked or unavailable, saved HTML and Mermaid remain available through file links. Visual verification is then reported as unavailable.
  40. - For `build-figma-library`, a connected Figma account and the official Figma plugin with the companion skills named in its instructions. Publishing and use across files depend on the account's Figma plan and file permissions.
  41. Miro and FigJam are optional integrations. Local discovery does not require an external board.
  42. ### Product Discovery
  43. These skills help you understand customers, examine product and market evidence, choose a problem worth solving, and decide what to test next. Use a skill on its own for a focused task, or use Supersprint to connect the work into a discovery cycle with saved decisions and a growing evidence tree.
  44. | Skill | What it does |
  45. |---|---|
  46. | `supersentence` | Clarifies a business idea through a compact interview and market research, ending with a super sentence, an evidence-based verdict and a next test. |
  47. | `supersprint` | Guides a discovery cycle with the included specialists, saved decisions and a growing HTML/Mermaid tree. |
  48. | `pres-prioritization` | Compares opportunities using Pain, Reach, Evidence and Strategic fit, with sourced scores and human review. |
  49. | `experiment-design` | Turns selected ideas into concrete prototypes, demand tests or commitment tests, with hypotheses and observable changes. |
  50. | [`ab-test-design`](plugins/product-discovery/skills/product-discovery/ab-test-design/SKILL.md) | Helps decide whether an A/B test is worth running, checks sample size and duration, and creates a simple offline HTML test card. |
  51. | `customer-research` | Finds public customer opinions and delivers a short report, sourced CSV, and HTML with grouped quotes and mention counts. |
  52. | `affinity-mapping` | Turns interview transcripts into evidenced themes, opportunities and an interactive evidence map. |
  53. | `desk-research` | Answers a research question using current sources, counterevidence and a separate evidence register. |
  54. | `interview-plan` | Creates research questions, a screener, an interview guide and a note-taking format. |
  55. | `interview-simulator` | Lets you practice customer interviews with a fictional persona, then evaluates your questions, technique and mistakes. |
  56. | `data-analysis` | Guides analysis from a business question through data checks, exploration and recommendations. |
  57. | `code-event-storming` | Reconstructs business processes from an existing codebase in an interactive EventStorming board with source excerpts and open questions. |
  58. #### A/B Test Design
  59. `ab-test-design` turns a proposed product change into a hypothesis, population, success metric and guardrail monitoring rule. It checks whether the available traffic can detect a useful effect, plans recruitment in full weeks with a 14-day minimum, and adds the outcome window before the final readout. The offline card keeps the experiment description on the left and the recommendation and one interactive calculator on the right. Guardrails use a simple predeclared check for detected deterioration; no detected drop is not proof of zero harm.
  60. #### Supersprint: from signals to a decision
  61. `supersprint` guides the process using the discovery skills above. It first establishes the product with you and asks about a missing outcome before doing analysis. "Not known yet" is a valid answer: the skill helps draft a working goal for your review. Briefs, CSV files, transcripts and previous research are optional inputs.
  62. ```text
  63. Goal -> Signals -> Opportunities -> Ideas -> Experiments -> Observed results -> Decisions
  64. ```
  65. At the Opportunities stage, Supersprint uses the included `pres-prioritization` skill before you choose a focus. It sums supported Pain, Reach, Evidence and Strategic fit scores, keeps missing evidence visible, and asks you to review the comparison and select an opportunity. The ranking does not select a winner automatically.
  66. The tree grows as you work, with an HTML view, a Mermaid export, highlighted selections and links to evidence. Product context and the current outcome stay visible above the tree, even before any branches exist. The context is saved in the project and mirrored to a native Canvas when the host supports that working surface.
  67. Each branch is centered on its own parent, with room between cards and zoom controls from 20% to 150%, reset and fit-entire-tree. Clicking a card highlights its connections; dotted lines from supporting signals appear only when you inspect the card they support.
  68. Titles and previews explain the actual person, situation and proposed change in natural language across Supersprint and its specialists.
  69. Each cycle saves its documents, decisions and state under `supersprint/<cycle>/` in your project, so you can resume later. No personal knowledge vault or external board is required.
  70. Before each stage, Supersprint asks what you want to add and waits for your reply. It then prepares the stage, shows the draft and updated tree, and waits for your review before moving on. Corrections return to the draft; a later conversation resumes the saved checkpoint. Explicit delegation can change this default. Tests state a hypothesis, practical method and expected observable change; you judge what counts as success. Plans and actual results stay separate.
  71. > Start a Supersprint for this product. Investigate why customers do not return, using the attached brief, CSV files and transcripts.
  72. ### Product Design
  73. | Skill | What it does |
  74. |---|---|
  75. | `super-landing-page` | Builds a distinctive landing page through a guided interview, independent art direction, a detailed execution prompt, and visual iteration. |
  76. | `extract-design-md` | Extracts measured styles and visual evidence from websites into a `DESIGN.md` for matching prototypes. |
  77. | `poster-people-ui` | Builds screens with crisp black outlines, paper surfaces, pink accents, and hand-drawn people. |
  78. | `visual-direction-research` | Researches existing websites, creates a comparative HTML moodboard with screenshots and evidence, then develops a detailed direction study after you choose references or ingredients. |
  79. | `document-design-system-components` | Turns visual documentation of an existing interface into component documentation for designers: anatomy, variants, properties, tokens, spacing, typography, usage and accessibility. After review, it can prepare an optional agent implementation package with your approval. |
  80. | `build-figma-library` | Turns product documentation, code or existing UI into a native Figma token and component library, with composition examples, usable properties, and verified handoff within or across files. |
  81. `visual-direction-research` starts from a product brief and real websites. It compares visual mechanisms, information density and relevant screen types, and keeps measured values, visual estimates and proposed adaptations distinct. You receive a complete HTML moodboard, choose a direction or mix of ingredients, and then receive a deeper HTML study. It does not automatically implement a site or create a library.
  82. `document-design-system-components` starts from visual materials for an existing design. It documents the observed component families and states, marks estimates and missing evidence, and presents the full result for review without reading from or writing to Figma. Once you approve the documentation, it offers an optional package for agents: `design.md`, reusable styles or components, and working usage examples. Its bundled instructions and templates are in Polish; you can request another output language.
  83. `build-figma-library` handles the Figma stage when you need native, editable assets. It reconstructs the documented system as Variables, styles and components, checks variant properties and composition examples, and verifies that another agent can reuse the library. Use it with the official Figma plugin and the companion skills listed in its instructions.
  84. ### Engineering
  85. | Skill | What it does |
  86. |---|---|
  87. | `production-readiness` | Guides people through preparation, deployment verification and observation of real use, with separate stage scores, evidence tied to concrete environments, agreed local repairs and one offline HTML checklist. |
  88. `production-readiness` creates new schema-2 reports or resumes that format at `docs/production-readiness.html` in the analyzed project. Legacy reports are left untouched; use a different output path to create a new report alongside one. Requirements cover access, data protection, user journeys, delivery, monitoring and recovery. The main checklist follows the current run: dev checks before deployment, then production checks after deployment. Later checks and previous results stay in separate expandable sections; environment identity is available in details. Preview/staging is optional. The Polish interface explains the next action, who can take it and how to verify it. All state, the Geist Sans font and its license are embedded in that HTML; no companion JSON file or network connection is required.
  89. Each stage has its own percentage of applicable mandatory requirements supported by current evidence, not a safety score. Preparation can complete while later checks wait for deployment or real use. Evidence identifies the assessed sources, exact environment, deployment and configuration. Missing evidence, stale checks and accepted risk cannot turn a blocker green. After reviewing a proposed repair package, the agent can implement local fixes and recheck results. It does not deploy; live checks require separate authorization. Bundled tests cover report rendering, scoring, evidence validation and audit history.
  90. > Nie jestem osobą techniczną. Sprawdź ten projekt i poprowadź mnie krok po kroku przez przygotowanie do produkcji. Pokaż checklistę HTML, blokery i proponowane poprawki.
  91. The source folders are `plugins/product-discovery/skills/product-discovery/`, `plugins/product-discovery/skills/product-design/` and `plugins/product-discovery/skills/engineering/`. All three are explicitly declared in the plugin manifests.
  92. ## Installation
  93. If you use a desktop app, you can add the marketplace from the interface without cloning this repository or using a terminal.
  94. ### ChatGPT Desktop
  95. Add the marketplace directly from the app:
  96. 1. Open **Plugins** from the left sidebar.
  97. 2. Select **Add+** in the top-right corner.
  98. 3. Select **Add marketplace**.
  99. 4. Paste [https://github.com/superhero-tech/superhero-skills/](https://github.com/superhero-tech/superhero-skills/) and confirm.
  100. 5. The **Superhero.tech** marketplace is now available in Plugins.
  101. 6. Install **Superhero Skills**.
  102. ### Claude Desktop
  103. Add the marketplace directly from the app:
  104. 1. Select your username in the bottom-right corner, then select **Settings**.
  105. 2. Select **Plugins**.
  106. 3. Select **Add+** in the top-right corner.
  107. 4. Paste [https://github.com/superhero-tech/superhero-skills/](https://github.com/superhero-tech/superhero-skills/) and confirm.
  108. 5. The **Superhero.tech** marketplace is now available in Plugins.
  109. 6. Install **Superhero Skills**.
  110. ### Claude Code
  111. Run these commands inside Claude Code:
  112. ```text
  113. /plugin marketplace add superhero-tech/superhero-skills
  114. /plugin install product-discovery@superhero-skills
  115. ```
  116. If Claude Code asks you to reload the plugin, run `/reload-plugins` or start a new session.
  117. Invoke a skill explicitly, for example:
  118. ```text
  119. /product-discovery:supersentence
  120. /product-discovery:supersprint
  121. /product-discovery:pres-prioritization
  122. /product-discovery:experiment-design
  123. /product-discovery:ab-test-design
  124. /product-discovery:customer-research
  125. /product-discovery:affinity-mapping
  126. /product-discovery:desk-research
  127. /product-discovery:interview-plan
  128. /product-discovery:interview-simulator
  129. /product-discovery:data-analysis
  130. /product-discovery:code-event-storming
  131. /product-discovery:production-readiness
  132. /product-discovery:super-landing-page
  133. /product-discovery:visual-direction-research
  134. /product-discovery:document-design-system-components
  135. /product-discovery:build-figma-library
  136. ```
  137. ### Codex
  138. Run these commands in your shell:
  139. ```bash
  140. codex plugin marketplace add superhero-tech/superhero-skills
  141. codex plugin add product-discovery@superhero-skills
  142. ```
  143. Start a new Codex session after installation. Version `1.11.0` adds `visual-direction-research` and `document-design-system-components` alongside the existing `build-figma-library`; check that your installed plugin has updated to this version or later to use the complete design workflow. Version `1.14.0` adds `code-event-storming`; version `1.15.0` adds `production-readiness`. Version `1.16.0` adds `ab-test-design`. You can also open `/plugins` to browse the `Superhero Skills` marketplace.
  144. Invoke a skill explicitly, for example:
  145. ```text
  146. $product-discovery:supersentence
  147. $product-discovery:supersprint
  148. $product-discovery:pres-prioritization
  149. $product-discovery:experiment-design
  150. $product-discovery:ab-test-design
  151. $product-discovery:customer-research
  152. $product-discovery:affinity-mapping
  153. $product-discovery:desk-research
  154. $product-discovery:interview-plan
  155. $product-discovery:interview-simulator
  156. $product-discovery:data-analysis
  157. $product-discovery:code-event-storming
  158. $product-discovery:production-readiness
  159. $product-discovery:super-landing-page
  160. $product-discovery:visual-direction-research
  161. $product-discovery:document-design-system-components
  162. $product-discovery:build-figma-library
  163. ```
  164. ### Install individual skills
  165. To select individual skills and install them in Codex, Claude Code, OpenCode or another supported agent, use [`skills.sh`](https://skills.sh):
  166. ```bash
  167. npx skills@latest add superhero-tech/superhero-skills
  168. ```
  169. Add `--global` if you want the skills available across all your projects. If you use Claude Code, choose either the plugin or individual installation to avoid installing every skill twice.
  170. This repository is a standalone marketplace for ChatGPT, Claude and Codex. Users add it directly from GitHub, so it does not need to be submitted to the Anthropic or OpenAI public plugin directories. Organizations can still restrict third-party marketplaces through managed settings.
  171. ## What earns a place here
  172. Skills earn a place here when:
  173. - We have used it on real product work.
  174. - It has survived contact with practitioners shipping actual products.
  175. - It does one clear job and works with other skills.
  176. - Its hard rules, checks and failure modes are written down.
  177. - It keeps decisions with the human where judgment matters.
  178. Test results do not substitute for use on real product work.
  179. Everything in this repository is released under the [MIT License](LICENSE). You can copy the skills, modify them, use them commercially and make them your own.
  180. ## Who's behind this
  181. We are Wojtek Strzałkowski and Piotr Kacała, founders of [Superhero.tech](https://superhero.tech) and creators of [AI Product Heroes](https://aiproductheroes.pl).
  182. We build products and teach Product Builders. We are skeptical of AI hype because we use these tools every day and have to live with their output. We judge each skill by whether it helps someone understand the problem, make a sharper decision or ship something customers need.
  183. The [Superhero.tech newsletter](https://newsletter.superhero.tech) is where we share what we are testing, what failed and what gets added to the Product Builder OS next.

Ratings and comments

Sign in to rate this skill.

Sign in

Free to use. If it’s helped, pass it on.