Clear answers about visual DNA and the ZIP.

No. Site Stripper gathers structured observations from public pages. It records visible page details and formats them into reference files. The resulting pack can then support a separate AI-assisted rebuild workflow.

Site Stripper works with publicly accessible HTTP or HTTPS websites. It follows relevant same-site pages selected from the site's structure. Pages that require a sign-in are outside the analysis path.

Hosted runs can analyze up to 10 pages and run for up to 150 seconds. A run can stop earlier when a source page is slow, protected, or returns incomplete information. The report records coverage and limitations.

No. Screenshot files are included when captures succeed and fit the hosted limits. The screenshot manifest records capture status, and the written notes and structured data remain useful when some images are unavailable.

No. It creates a structured reference pack, not a working copy of the source site. It does not submit source-site forms, bypass bot checks, or replay arbitrary JavaScript interactions.

The package contains Markdown guides, page notes, JSON records, theme files, and screenshot files when available. It also includes a run summary and report data that describe the selected pages and coverage.

The sample pack shows a real file tree and excerpts from a generated package.

It starts from public site structure, including site-map entries, navigation, and relevant same-site links. Candidate pages are ranked for a bounded run. The selection report records which candidates were included and why.

Site Stripper analyzes pages that are publicly accessible. It does not sign in to source websites, submit their forms, or use private account sessions. If a page requires authentication, it will not be available as a normal public source page.

Start with README.md for the source URL, run summary, selected pages, and coverage. Then read REBUILD.md for the high-level build guidance and open the page-specific notes for source content and structure.

No. The package contains observations and reference material such as page notes, structured JSON, theme files, and available captures. It does not include the source site's HTML, CSS, or application code.

It is the combination of details that gives a site its recognizable look and feel: page structure, content hierarchy, palette, typography, spacing, layout, imagery, responsive behavior, and art direction. The reference ZIP gathers these observations into files a team or agent can use while building a new site.

Yes. The comprehensive ZIP includes readable Markdown, structured JSON, design tokens, layout and responsive notes, and assets and captures when available. Use it as the starting brief for Claude, Codex, OpenCode, Cursor, and any other AI agent that can read project files. Ask the agent to create a brand-new site with the reference's visual style, art direction, and overall feel.

One recorded one-prompt rebuild of a clean content site scored 93.2/100 using a Site Stripper pack. This is a measured result for that site and setup, not a guarantee for every website. Results vary with complexity, interactions, capture coverage, and the AI agent.

The pack provides detailed evidence to guide a close visual match, but it does not promise an exact copy. A result depends on the source pages available, the completeness of captures, the implementation choices, and the capabilities of the AI agent. Review the new site against the sample captures and notes as you build.