Site Stripper
Sign up

Product features

Capture a website's visual DNA in one pack.

Site Stripper turns page structure, visible content, design tokens, layout, responsive behavior, imagery, and available captures into a comprehensive reference ZIP. Use it to start a new site in the reference's style with Claude, Codex, OpenCode, Cursor, and any other AI agent that can read project files.

Create your account See a sample pack

Page discovery

Find likely key pages from public navigation, site-map entries, and relevant same-site links. The selection report preserves the candidates and explains which pages made it into a bounded run.

Structured observations

Keep headings, sections, visible copy, layout samples, control labels, media references, and design tokens connected to their source page.

Available captures

Review desktop, mobile, and section captures when they are available. Slow or protected pages and hosted limits can make capture coverage partial.

Machine-readable records

Use page, token, navigation, interaction, asset, selection, responsive, and report data alongside the written guides.

Rebuild documentation

Carry page notes, information architecture, responsive observations, typography, imagery, content style, and implementation guidance forward.

Portable ZIP delivery

The browser validates the generated documents and structured files, then assembles a portable ZIP with capture binaries when available.

A build brief for AI agents

Give a new website the same visual DNA.

The comprehensive ZIP brings page structure, visible content, design tokens, typography, layout data, responsive notes, art direction, and source assets and captures when available into one usable brief. Start a brand-new site with Claude, Codex, OpenCode, Cursor, and any other AI agent that can read project files. The pack helps it carry the reference site's visual language into a fresh build.

1 Source

Analyze a public website

Site Stripper records selected pages, content, navigation, design details, and available visual captures as evidence of the site's visual DNA.

2 ZIP

Provide the reference files

Page notes and structured records give the agent concrete source material.

  • REBUILD.md
  • data/tokens.json
  • data/layout.json
  • pages/*.md
  • assets/ and screenshots/ when available
3 Agent

Build from the pack

Use the ZIP or its extracted folder with Claude, Codex, OpenCode, Cursor, and any other AI agent that can inspect project files.

ClaudeCodexOpenCodeCursorAny other AI agent
4 Rebuild

Match the reference style

Use the extracted palette, typography, layout, imagery, interaction notes, and art direction to guide the new website.

93.2 / 100One recorded one-prompt evaluation on a clean content site scored 93.2/100. This result describes that site and setup. Results vary with site complexity, interactions, available evidence, and the AI agent.
Mobile homepage view of the sample Morrow House site, showing a stacked layout and project imagery
A narrow layout can change the order and scale of the same source content.

Responsive evidence

See how a page adapts to a smaller screen.

Responsive notes help a team compare page structure across viewports. The reference pack can include desktop and mobile captures when available, plus responsive observations that describe what changed.

  • Compare layout shifts and content order.
  • Review observed typography and spacing patterns.
  • Keep capture coverage visible in the report.
How the process works

Package structure

Start with the overview, then open page-level files.

The output separates the page map, individual page notes, reusable design observations, and structured records.

Start with the overview

README.md summarizes the source, page coverage, and report status. REBUILD.md gathers implementation guidance in one place.

Trace the page set

Information architecture and selection records show how the analyzed pages relate and why candidates were selected or excluded.

Inspect design evidence

Token, typography, responsive, imagery, and interaction notes capture details that a screenshot alone cannot explain.

Use the structured files

JSON records keep page, asset, navigation, layout, and report details available for tools that prefer structured input.

Run limits

Review coverage and capture status.

Hosted analysis can return partial observations when pages are protected, slow, or media-heavy.

  • Up to 10 pagesMaximum selected pages in one run.
  • 150 secondsHosted analysis wall limit.
  • Captures when availableSource-site behavior and hosted limits affect screenshots.

Site Stripper observes public pages. It does not submit source-site forms, solve CAPTCHA challenges, bypass bot checks, or replay arbitrary JavaScript interactions.

Use the same source notes across the team.

Designers can compare page patterns. Developers can use the page map and content notes. AI-assisted rebuild workflows can read the Markdown and JSON files as additional source context.

Design reviewDiscuss structure, hierarchy, responsive behavior, and visual tokens using shared source notes.
ImplementationFind page content, repeated patterns, and available visual captures in the same package.
Rebuild workflowsUse Markdown and JSON references with the tools your team already uses.

Explore the output

See how the files fit together.

Open a representative file tree and inspect the kinds of documentation generated from a real analysis.

Browse the sample pack