Guide · Agent skill
How to use the URL to Code agent skill: install and clone a page
Updated · written with AI from the project's own documentation and checked against our catalogue data
URL to Code is an agent skill from the MiMo-Code project that tells an AI agent how to recreate a live website or app URL as a runnable, frontend-only local build. It is for people who have permission to reproduce a page and want a close local reference. After this guide you will know what the skill does, how to install it, how to ask for it in plain language, and what limits and safety checks apply.
Quick start
- 1Confirm permission. Make sure you own the page or have permission to reproduce it, and that you have checked the site's terms.
- 2Download and unzip. Download the url-to-code ZIP from our page and unzip it.
- 3Place the skill folder. Put it in ~/.claude/skills/ for personal use or in your project's .claude/skills/ folder, or upload it under Settings → Capabilities → Skills in the Claude apps.
- 4Ask in plain language. Request a faithful recreation of the URL and state that you have permission to reproduce it.
- 5Let the agent capture the page. The agent opens the page, captures desktop and mobile views and inspects structure, styles and interactions.
- 6Review the build and QA report. Run the local implementation, compare it with the original and read design-qa.md before accepting the result.
What an agent skill is and what url-to-code makes the agent do
An agent skill is a folder with a SKILL.md file that gives an AI agent a set of instructions for a specific kind of task. When a request matches the skill, the agent follows its workflow instead of improvising.
The url-to-code skill covers one job: recreating a live page you own or have permission to reproduce. The agent first opens the page in a browser, captures desktop and mobile views, and inspects the structure, styles and interactions. It saves permitted images and assets locally, builds a frontend-only version, compares that build side by side with the original, and writes a design QA report before handing the work back.
- Reminds you to check permissions and the target site's terms first
- Stops if the browser lands on a wrong page, login wall, error or unrelated redirect
- Does not hotlink source assets in the final result
- Uses your repository's existing framework; React is not required
Install the url-to-code skill
Download the ZIP for the skill from our page and unzip it. You can install it in one of two places, depending on whether you want it for all projects or just one.
For personal use across projects, unzip it into ~/.claude/skills/. For a single project, unzip it into that project's .claude/skills/ folder. In the Claude apps, you can instead upload the skill under Settings → Capabilities → Skills.
- Personal install: ~/.claude/skills/
- Project install: .claude/skills/ inside your project
- Claude apps: Settings → Capabilities → Skills, then upload
Trigger the skill with a natural request
You do not need a special command. Ask the agent in plain language to clone or faithfully recreate a page, and state that you own it or have permission. The skill is meant for faithful reproduction, so say that explicitly.
If you want a redesign or a page loosely inspired by another site, say so. The skill routes those requests to a different workflow for context and ideation rather than cloning.
- Good: "Recreate https://example.com/landing as a local frontend. I own this site."
- Good: "Clone our legacy marketing page into the new repo. We have permission to copy it."
- Not this skill: "Make a modern redesign inspired by this page."
What the skill needs before it runs
The agent needs a browser it can drive to load and inspect the live page. The skill also refers to three files inside the same bundle: ../index/SKILL.md, ../design-qa/SKILL.md and ../../references/critical-overrides.md. Keep the bundle intact so those references resolve.
The source files name no API keys or accounts. The skill runs scripts and uses the network, so review what your agent is doing on your machine.
- A browser the agent can control
- The full bundle, including the referenced SKILL.md files
- Network access to load the live page
Safety notes on scripts, network and permissions
The skill runs scripts and uses the network. It loads the page you point it at and saves permitted assets to your machine, so only use it on URLs you are allowed to reproduce.
The skill reminds you about copyright and terms of service but does not enforce them. Copying someone else's site without permission can raise copyright and terms-of-service problems. The skill is also limited to frontend-only output, so it will not build backend features or work past login walls.
The skill also says not to publish unless you ask and the host exposes an authorized deployment tool.
- Check permission and the site's terms before you start
- Expect the agent to stop at login walls
- Frontend-only output: no backend behavior is built
- Deployment happens only when you ask and a tool is authorized
Tips for better results
Give the agent a clear target: the URL, the viewport concerns and any pages or states that matter most. The skill tests interactions one at a time and returns to the starting state between tests, so name the interactions you care about.
Let the agent finish its design QA. The skill asks for design-qa.md and only hands off after a passed result or a documented verification blocker. Where it substitutes an asset, it discloses the substitution, so read that note.
- Name the URL and the key interactions up front
- Ask for desktop and mobile views to be compared
- Read design-qa.md before accepting the build
- Expect substitutions to be disclosed when an asset cannot be saved
What people use it for
- Recreating a client's existing marketing page in a new local frontend codebase
- Migrating a legacy landing page into a modern project for side-by-side comparison
- Producing a pixel-close local reference of a page you are authorized to reproduce
- Checking responsive layouts by comparing desktop and mobile captures of the local build
Frequently asked questions
How do I install the url-to-code skill?
Download the ZIP from our page and unzip it into ~/.claude/skills/ for personal use, or into a project's .claude/skills/ folder. In the Claude apps you can instead upload it under Settings → Capabilities → Skills.
What does the url-to-code skill require?
It requires a browser the agent can drive to load and inspect the live page, plus the referenced files ../index/SKILL.md, ../design-qa/SKILL.md and ../../references/critical-overrides.md in the same bundle. The source names no API keys or accounts.
Is url-to-code free and what is its licence?
The source project is licensed under MIT. The source files do not mention pricing, so check the project's repository for any cost details.
Can I use url-to-code to copy any website?
No. The skill is for pages you own or have permission to reproduce. It reminds you to check permissions and terms, but it does not enforce them, so the responsibility is yours.
Why does the skill stop on some pages?
It stops if the browser resolves to a wrong page, login wall, error, unrelated redirect or another surface that cannot ground the clone. It is also limited to frontend-only output, so sites that depend on backend features are not a good fit.
What should I use instead for a redesign?
For redesigns, improvements or loosely inspired pages, the skill routes you through ../index/SKILL.md to context and ideation rather than faithful cloning.
New agent skills, repos and AI tools — every week
The best new skills and open-source projects, the fastest-growing repos, and one free prompt. Five minutes to read. Unsubscribe anytime.
Our verdict
Use url-to-code if you have permission to reproduce a live page and want a frontend-only local build checked side by side against the original. Pick a different approach if you need a redesign, a page behind a login, or anything that depends on backend features, or if you have no rights to the source site.
Download the skill