# url2prompt

## Summary

url2prompt is a web application that converts a public website URL into a professional English prompt for rebuilding an original web experience with an AI coding agent.

The user enters a URL such as `vibecode.how`, selects an analysis mode, and receives a structured prompt describing what an implementation agent should build. The output is a build instruction, not a copy of the source website.

## Primary Use Case

Use url2prompt to:

- understand the visual language of a website;
- reproduce a similar product experience with original content and assets;
- communicate layout and interaction requirements to an AI coding agent;
- turn a visual reference into a concrete frontend implementation brief.

The generated prompt should preserve useful design principles while avoiding copied proprietary text, logos, illustrations, and brand assets.

## Workflow

1. Accept a public HTTP or HTTPS website URL.
2. Normalize common input such as a hostname without a scheme.
3. Reject local, private, or unsafe network destinations.
4. Extract design evidence from the rendered website.
5. Generate one English implementation prompt from that evidence.
6. Stream the prompt to the browser as it is generated.
7. Let the user copy the completed prompt for use in an AI coding agent.

## Analysis Modes

### Fast mode

Fast mode is intended for routine analysis and higher-throughput usage. It extracts useful design evidence quickly, including page structure, visual tokens, typography, spacing, components, and responsive observations where available.

### Pro mode

Pro mode is intended for important or complex pages. It performs a deeper design-system-oriented extraction and can provide richer information about tokens, layout patterns, component states, and implementation details.

## Prompt Output

The generated prompt is written in English and organized for direct use by a senior frontend engineer or AI coding agent. It should cover:

- product intent and page goals;
- information architecture and page hierarchy;
- visual system, including colors, typography, spacing, radii, borders, and elevation;
- components and reusable patterns;
- desktop, tablet, and mobile behavior;
- interactions and states such as hover, focus, active, loading, and error;
- accessibility requirements;
- implementation constraints and suggested frontend architecture;
- originality requirements for content, branding, and assets.

The prompt should distinguish observed evidence from assumptions. Missing details should be stated as unknown rather than invented.

## Privacy and Sessions

No account is required for the public workflow. Anonymous browser sessions are isolated from one another. Analysis records are associated with a session token and are not intended to be shared between users.

## Public Interface

- Homepage: https://url2prompt.com/
- Sitemap: https://url2prompt.com/sitemap.xml
- Short LLM index: https://url2prompt.com/llms.txt

## Limitations

- The input website must be publicly reachable.
- Extraction quality depends on the target site's accessibility, JavaScript behavior, network response, and rendered state.
- A generated prompt is an informed implementation brief, not a guarantee of pixel-perfect source reconstruction.
- The service should not be used to copy protected brand identity, proprietary text, or private content.

## Recommended Interpretation

Treat the output as a high-signal frontend specification. Preserve the source page's useful hierarchy and interaction ideas, then create an original implementation with accessible semantics, responsive behavior, and maintainable components.