Public URL Content Capture

Capture and verify a public webpage as Markdown or JSON.

When to use it

Save a public page

I have an HTTP or HTTPS page that opens without a login and want its rendered main content saved locally, with optional media download.

Extract a transcript

I need the public transcript from a YouTube video, or the readable content from an anonymously accessible X post or thread.

Capture a discussion

I want a Hacker News discussion or another public page captured as Markdown or structured JSON for later use.

Won't take

This is not for pages that require login, CAPTCHA, payment, or manual browser action, and it does not summarize, rewrite, translate, or publish the captured content.

What it produces

A successful command is still provisional until the saved title and body are inspected.

• Content file: Saves clean Markdown or JSON at a conflict-safe path grouped by domain and page
• Media files: Optionally downloads images and videos beside the capture and rewrites local links
• Capture report: Returns the output path, selected adapter, media result, and any unverified limitation
• Preferences: On first use, asks where to save captures and how to handle media before writing EXTEND.md
• Browser activity: Uses Chrome through the bundled command and may reuse an explicitly supplied CDP endpoint
• Never: Exports authentication state, asks the user to defeat an access wall, or preserves an unusable login page as success

Prerequisites & boundaries

Prerequisites

Provide exactly one public HTTP or HTTPS URL. The environment needs Bun, Chrome or Chromium, writable output storage, and first-use preference choices when no EXTEND.md exists.

Neighbor skill split

• Saving one public page unchanged → url-to-markdown
• Researching and verifying claims across sources → collect-sources

Scenarios it declines

• Login walls, CAPTCHA, Cloudflare verification, or any manual interaction
• Multiple ambiguous URLs or an unsupported forced adapter
• Requests to summarize, edit, translate, or publish the captured page

Subtle edges

• Site adapters cover X, YouTube, Hacker News, and generic public pages
• Media behavior follows explicit arguments first, saved preferences second, and defaults last
• A low-quality capture is removed and reported as unsupported even when the command exits successfully