mcpware

Pagecast

Community mcpware
Updated

Record any browser page as GIF or video via MCP — powered by Playwright + ffmpeg

Pagecast

npm versionlicenseGitHub starsGitHub forks

English | 廣東話

The MCP server that turns agent-run browser sessions into GIFs and videos you can actually ship.

Tell your AI to record what it does. It opens a browser, does the thing, and hands you back a GIF, WebM, or MP4. You watch it happen live.

The demo below was recorded by Pagecast itself. Yes, really.

Moltbook Demo

Why

You built a web app. You need a demo GIF for the README. So you:

  1. Open a screen recorder
  2. Manually click through the demo
  3. Export, crop, optimize
  4. UI changes → repeat everything

Or — you tell your AI:

"Go to https://myapp.localhost:3000, play around with it for 5 seconds and make a GIF"

The AI calls record_and_gif → outputs recordings/recording-{id}.gif. Done.

The browser is visible by default — you watch the AI work in real time.

Use Cases

What you say What happens Output
"Record my app and make a demo GIF" AI opens browser, interacts, exports GIF .gif
"Log into dashboard, toggle dark mode, record it" AI performs full workflow, captures everything .webm .mp4
"Submit an empty form, record what errors show up" AI does QA, you review the recording later .webm
"Walk through the signup flow and record a video" AI creates onboarding documentation .mp4
"Post on Moltbook and record yourself doing it" AI records its own marketing material 🤯 .gif
"Record my app in 1080x1920 for an IG Reel" AI records in vertical format for social media .mp4

📐 Any size. Any platform.

Record in any viewport size — the AI opens a real browser at whatever dimension you need:

Format Size For
1280×720 16:9 GitHub README, YouTube, docs
1080×1920 9:16 IG Reels, TikTok, YouTube Shorts
1080×1080 1:1 Instagram posts, LinkedIn
Custom Any Whatever you need

Quick Start

Node.js ≥ 20 and ffmpeg required.

# Add to Claude Code
claude mcp add pagecast -- npx -y @mcpware/pagecast

# Or run directly
npx @mcpware/pagecast

# Headless mode (no visible browser)
claude mcp add pagecast -- npx -y @mcpware/pagecast --headless

# First time: install browser
npx playwright install chromium

MCP Tools

Tool What it does
record_page Open a URL in Chromium, start recording. Returns session ID
interact_page scroll, click, hover, type, press keys, select dropdowns, navigate — all captured on video
stop_recording Stop and save as .webm
convert_to_gif WebM → optimized GIF (two-pass palette, configurable FPS/width/trim)
convert_to_mp4 WebM → MP4 (H.264, widely compatible for social/sharing/embedding)
record_and_gif All-in-one: record URL for N seconds → GIF
list_recordings List all .webm, .gif, and .mp4 files in output directory

How It Works

AI → MCP tools → Playwright (headed browser + video capture)
                       ↓
                  .webm recording
                       ↓
              ffmpeg two-pass palette
                       ↓
               optimized .gif / .mp4
  1. record_page launches Chromium with recordVideo enabled
  2. interact_page performs actions — scroll, click, hover, type, press keys
  3. stop_recording flushes the video to disk
  4. convert_to_gif runs two-pass ffmpeg: first extracts an optimal 256-color palette from all frames, then encodes with Bayer dithering. Way better quality than single-pass.

Comparison

Tool Automated Interactions Output AI-driven
Pagecast ✅ click/type/scroll/hover GIF + WebM + MP4
gifcap.dev GIF
Peek / ScreenToGif / Kap GIF
Playwright MCP (official) Screenshot only Partial
playwright-record-mcp WebM only Partial
VHS (Charmbracelet) Terminal only GIF

Pagecast is the only MCP that does browser recording + AI interactions + GIF/video output.

Configuration

Setting Default Notes
Browser Headed (visible) --headless for background
GIF FPS 10 Higher = smoother, larger
GIF width 640px Height auto-scaled
Video viewport 1280×720 Downscaled for GIF
Output dir ./recordings Override: RECORDING_OUTPUT_DIR

Architecture

src/
├── index.js       # MCP server — 7 tools, stdio transport
├── recorder.js    # Playwright browser lifecycle + sessions
└── converter.js   # ffmpeg two-pass GIF conversion
  • Headed by default — watch what the AI does
  • Lazy browser — Chromium only launches on first recording
  • Session-based — multiple simultaneous recordings
  • One browser, multiple contexts — each recording is isolated
  • execFile not exec — safe against shell injection

More from @mcpware

Project What it does Install
UI Annotator Hover any element to see its name — zero extensions, any browser npx @mcpware/ui-annotator
Claude Code Organizer Visual dashboard for memories, skills, MCP servers, hooks npx @mcpware/claude-code-organizer
Instagram MCP 23 tools for the Instagram Graph API npx @mcpware/instagram-mcp

License

MIT

MCP Server · Populars

MCP Server · New

    mcpware

    Cross-Code Organizer (CCO)

    Cross-Code Organizer (formerly Claude Code Organizer): cross-harness config dashboard for Claude Code, Codex CLI, MCP servers, skills, memories, agents, sessions, security scanning, context budget, and backups.

    Community mcpware
    dcostenco

    🧠 Prism MCP — The Mind Palace for AI Agents

    The Mind Palace for AI Agents - HIPAA-hardened Cognitive Architecture with on-device LLM (prism-coder:7b), Hebbian learning, ACT-R spreading activation, adversarial evaluation, persistent memory, multi-agent Hivemind and visual dashboard. Zero API keys required.

    Community dcostenco
    Mcp-Brasil

    mcp-brasil

    MCP Server para 70 APIs públicas brasileiras

    Community Mcp-Brasil
    kridaydave

    File Organizer MCP Server

    This MCP server will organize your files using connections to MCP using clients like Claude, Cursor and Gemini Cli

    Community kridaydave
    higress-group

    AI Gateway

    🤖 AI Gateway | AI Native API Gateway

    Community higress-group