Skip to content

Repository files navigation

Product-Skills

Turn Product Team requests into verified working changes with AI Coding.

YOU DESCRIBE WHAT.
AI CODING HANDLES HOW.
YOU REVIEW THE WORKING RESULT.

Product Team describes a request, AI Coding builds and verifies, then a working Preview is reviewed

Who is this for? Product Team.

What is it? A Cursor Plugin that helps AI Coding analyze your existing project, implement the change you asked for, verify it, and prepare a Preview when supported.

You do not need to understand skills, agents, hooks, manifests, or repository internals.


Start here

1. Install Product-Skills in Cursor

Primary path: install the product-skills plugin from the Cursor Marketplace.

Status: Ready for Cursor Marketplace submission.
This repository is not claiming Marketplace publication until Cursor approves and lists it.

Until Cursor publishes the listing, maintainers may load the plugin locally for testing (see Advanced).

2. Open your existing project

Open the application you want to change — not this repository.

3. Tell AI Coding what you want

Use natural language. Example:

Refactor this existing project to React + TypeScript + Feature-Sliced Design.
Do not change product behavior.
Verify everything and deploy a Vercel Preview for review.

Or run the Cursor command react-fsd-refactor.

4. Review the working result

Primary review surface: the Vercel Preview (when the project is configured and access exists).

If Preview cannot be created, AI must explain the blocker and still complete local verification that is possible.


How it works

Five journey stages from request through build, verify, review, and iterate

You describe WHAT.
AI determines HOW.


Example: React + TypeScript + FSD migration

You have an existing application.
You want better frontend architecture without redesigning the product.

Existing app inspected, migrated to React TypeScript FSD, verified, then reviewed on Preview

Copy this prompt

Refactor this existing project to React + TypeScript +
Feature-Sliced Design (FSD).

This is an architecture refactoring only.

Preserve:

- existing product behavior
- business rules
- user workflows
- permissions
- API behavior
- data behavior
- existing UX

Do not:

- add new features
- remove existing features
- redesign the product
- change business behavior
- rewrite unrelated code

First inspect the existing project.

Determine the appropriate FSD boundaries from the actual
project.

Implement the migration.

Verify:

- FSD architecture
- typecheck
- lint
- tests when available
- production build
- important runtime journeys

Fix all verification failures.

If Vercel Preview is supported and access exists,
deploy and verify it.

Return:

- summary of changes
- verification status
- Preview URL

Review the result

Feedback loop centered on Product Review

Check

Question
Does the requested change work?
Does the existing workflow still work?
Is anything missing?
Did anything unexpected change?
Is the user experience correct?

You do not need to inspect source code, Git diffs, FSD folders, skills, or AI configuration.

About Vercel

If the project is configured for Vercel and the AI Coding environment has the required access, AI can deploy and verify a Preview.

Product-Skills does not always deploy to Vercel. If deployment cannot run, AI must report the blocker — not invent a URL.


More prompts

Add a feature

I want to add this feature:

[describe feature]

Users: [users]
Goal: [goal]

Keep existing behavior unless I explicitly change it.
Analyze first. Implement. Verify. Provide a Preview when supported.

Fix a bug

There is a problem:

[describe problem]

Expected: [expected]
Actual: [actual]

Investigate first. Make the smallest safe fix. Verify. Provide a Preview when supported.

Improve UX

I want to improve this user experience.

Current user problem: [problem]
Goal: [goal]

Keep unrelated business behavior unchanged.
Review the current flow first. Implement. Verify. Provide a Preview when supported.

What Product Team provides

Product Team owns What Why Constraints Feedback; AI Coding owns How

Product Team AI Coding
What you want Project analysis
Why you want it Technical approach
What must not change Implementation
Expected result Verification
Feedback Fixes + new Preview

FAQ

What project do I open?
Your existing application.

Do I need to learn skills or agents?
No.

Can I ask for React + TypeScript + FSD?
Yes. Use the example above or the react-fsd-refactor command.

Where do I review?
The working Preview / Staging application.

Does Product-Skills require Vercel, GitHub, or Supabase?
No. Those are optional when a specific task needs them.


Advanced — plugin maintainers / Marketplace testing

Not for Product Team daily use.

Local Cursor Plugin test (required before submission)

Official docs: Test plugins locally

Windows path

%USERPROFILE%\.cursor\plugins\local\product-skills\

Steps

  1. From this repository run:
npm run install:local-plugin

Or manually copy the repo to %USERPROFILE%\.cursor\plugins\local\product-skills\ (plugin root must contain .cursor-plugin/plugin.json). 2. In Cursor: enable Include third-party Plugins, Skills, and other configs if present. 3. Run Developer: Reload Window. 4. Open Settings → Plugins (or Customize → Plugins). 5. Confirm plugin name: product-skills. 6. Confirm components load:

  • skills (core Product / Delivery / Quality)
  • rules
  • agents (debugger, explorer, reviewer, verifier)
  • commands (product-request, react-fsd-refactor)
  1. Open any app project and try the command react-fsd-refactor (or paste the README FSD prompt).

If the plugin does not appear: confirm the path is plugins\local\ (not plugins\cache\local\), then Reload Window again.

Maintainer validation

npm run validate

Optional project bootstrap scripts

install/ helpers copy guidance into another repository for non-Marketplace workflows. They are maintainer utilities, not the Product Team install path.

  • node install/bootstrap.mjs --target <project>
  • node install/install.mjs --target <project>

Official source: https://github.com/irwebtools/Product-Skills

See SECURITY.md and install/README.md.


License

MIT

About

Lightweight AI coding harness for PMs and BAs to turn product requests into verified React changes that developers can continue toward production.

Topics

Resources

Security policy

Stars

0 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages