Pagebolt
Capture live pages as screenshots, PDFs, OG images, or narrated videos and inspect markup without leaving Claude Code or Cursor.
Overview
PageBolt is a MCP server for the Build phase that captures screenshots, PDFs, OG images, page inspections, and narrated videos from URLs for Claude and Cursor.
What is this MCP server?
- Full-page and viewport screenshots via MCP tools against real URLs
- PDF export and Open Graph image generation for share-ready previews
- Structured page inspection for titles, meta, and layout signals
- Narrated video recording of browser sessions for demos and bug reports
- stdio npm package pagebolt-mcp with PAGEBOLT_API_KEY for Claude and Cursor
- Registry server version 1.8.2 with npm package identifier pagebolt-mcp (published package version 1.8.1 in schema)
- One required secret env var: PAGEBOLT_API_KEY
- stdio transport MCP server hosted at github.com/Custodia-Admin/pagebolt-mcp
Community signal: 2 GitHub stars.
What problem does it solve?
Checking how a real page looks, exports, and shares—without leaving your agent or manually driving a browser—slows solo builders during UI and launch prep.
Who is it for?
Solo builders validating staging URLs, OG cards, and walkthrough videos while pair-programming with Claude Code or Cursor.
Skip if: Teams that need a full managed browser farm, deep security crawling, or design-only mockups with no live URL.
What do I get? / Deliverables
After you register pagebolt-mcp and set PAGEBOLT_API_KEY, your agent can pull visual assets and inspection output from URLs in the same workflow as your code.
- PNG or equivalent screenshots and PDF exports of target URLs
- Open Graph images sized for social link previews
- Page inspection output and narrated video recordings of browsed sessions
Recommended MCP Servers
Journey fit
Solo builders most often wire PageBolt while implementing and checking UIs, so the canonical shelf is Build rather than a single launch-only utility. Page capture, inspection, and social-preview assets map directly to frontend verification and documentation while pages are still in flux.
How it compares
Browser-capture MCP integration, not an agent skill or static screenshot CLI with no page inspection.
Common Questions / FAQ
Who is PageBolt for?
PageBolt is for solo and indie builders using Claude Code or Cursor who want screenshots, PDFs, OG images, inspections, and narrated recordings from real web pages via MCP.
When should I use PageBolt?
Use PageBolt when you need production-faithful visuals or meta checks from a URL during frontend work, launch assets, or demo videos without switching to a separate browser toolchain.
How do I add PageBolt to my agent?
Install the npm MCP server pagebolt-mcp, add it to your Claude Code or Cursor MCP config with stdio transport, and set the required secret environment variable PAGEBOLT_API_KEY from https://pagebolt.dev.