The capture API built for AI agents and developers.
Screenshots. PDFs. Demo videos. Browser automation. One API — usable by humans and AI agents.
PageBolt is a REST API for capturing the web. But unlike traditional screenshot tools, it ships with a built-in MCP server — so AI coding assistants like Cursor, Claude Desktop, and Windsurf can capture screenshots and record demo videos as part of their workflow. No glue code. No plugins.
What makes it different:
- AI-Native (MCP Server) — Your AI agent can call PageBolt directly. Ask Cursor to "record a demo of this feature" and it happens. Screenshots, videos, PDFs — all from your AI assistant's context.
- Agentic CI/CD Demos (New) — A GitHub Action inspects your deploy previews and auto-generates demo videos for every pull request. Posted as PR comments. Reviewers see the feature, not just the diff.
- AI Voice Narration — Demo videos come with optional text-to-speech narration synced to each step. 10+ professional voices via Azure AI Speech and OpenAI TTS. Your agent writes the script, PageBolt speaks it.
The full toolkit:
- Demo Video Recording — MP4/WebM/GIF with cursor highlighting, click effects, step tooltips, macOS/Windows frames, and gradient backgrounds.
- Screenshot API — Pixel-perfect PNG/JPEG/WebP. 30+ parameters: device emulation, ad/cookie/chat blocking, full-page scroll, dark mode, styled output with one-click themes.
- PDF Generation — Any URL or HTML to PDF. Custom margins, headers/footers, page ranges.
- OG Image Generation — Social cards from templates or custom HTML.
- Browser Automation — Chain clicks, fills, navigation, and captures into multi-step sequences.
Who it's for:
Developers using AI coding agents. Teams that want every PR to ship with a video demo. PMs and marketers who need visual assets without wrangling Puppeteer. Anyone building AI-powered workflows that need to see and capture the web.
The problem:
Screenshot APIs are a solved problem. But the world moved to AI agents — and agents can't use your Puppeteer scripts. They need a protocol-native API they can call directly. Meanwhile, demo videos still require manual screen recording or fragile toolchain setups.
PageBolt is the capture layer purpose-built for this new workflow: one API that both your code and your AI agent can call.
Why it matters:
- Agent-ready — MCP server means your AI assistant captures, records, and automates without leaving the editor.
- Demos on autopilot — Every PR gets a video walkthrough, posted automatically.
- Kill complexity — One API replaces Puppeteer + FFmpeg + TTS + screenshot libraries.
- Ship faster — Stop building capture infrastructure. Start shipping product.