MCP Scroll Animation Analyzer
Analyzes scroll-triggered animations on any webpage using Playwright, extracting GSAP, ScrollTrigger, Lenis, and CSS code via AST and runtime detection.
README
MCP Scroll Animation Analyzer
A Model Context Protocol (MCP) server that uses Playwright to analyze scroll-triggered animations on any webpage. Unlike GSAP generation MCPs, this focuses on analyzing existing websites to extract and understand animations built with GSAP, ScrollTrigger, Lenis, and CSS.
Features
- AST-based extraction - Uses Acorn for reliable JavaScript parsing (not regex)
- Open any webpage in a real Chromium browser
- Scroll and record - capture video and screenshots as you scroll
- Extract animation code - find GSAP, ScrollTrigger, and Lenis code
- Interactive analysis - click, hover, and interact while recording
- Watch specific elements - track style changes on individual elements
- Generate reports - JSON, Markdown, or HTML reports
- Export replicable code - generate standalone code (vanilla JS, React, Vue)
- CLI support - use standalone without MCP integration
Installation
1. Install dependencies
cd mcp-scroll-analyzer
npm install
2. Install Playwright browsers
npx playwright install chromium
3. Build the TypeScript
npm run build
4. Configure Claude Code
Add the MCP server to your Claude Code configuration.
For VSCode Claude Code extension, add to your .vscode/mcp.json:
{
"mcpServers": {
"scroll-analyzer": {
"command": "node",
"args": ["C:/path/to/mcp-scroll-analyzer/dist/index.js"]
}
}
}
For Claude Code CLI, add to ~/.claude/claude_desktop_config.json:
{
"mcpServers": {
"scroll-analyzer": {
"command": "node",
"args": ["C:/path/to/mcp-scroll-analyzer/dist/index.js"]
}
}
}
CLI Usage
The analyzer can also be used standalone via CLI:
# Analyze a webpage
npx scroll-analyzer analyze https://example.com
# With options
npx scroll-analyzer analyze https://example.com --output ./reports --format html
# Export replicable code
npx scroll-analyzer export https://example.com --target react
CLI Commands
analyze <url> - Analyze a webpage and generate a report
Options:
-o, --output <dir> Output directory (default: ./scroll-analysis-output)
-f, --format <type> Report format: json|markdown|html (default: markdown)
--no-headless Run browser in visible mode
-w, --width <pixels> Viewport width (default: 1920)
-h, --height <pixels> Viewport height (default: 1080)
export <url> - Analyze and export replicable code
Options:
-o, --output <dir> Output directory
-t, --target <fw> Framework: vanilla|react|vue (default: vanilla)
Available MCP Tools
Page Management
| Tool | Description |
|---|---|
open_page |
Open a URL in headless browser for analysis |
close_browser |
Close browser and clean up resources |
set_output_directory |
Set where screenshots/videos/reports are saved |
Scroll & Recording
| Tool | Description |
|---|---|
scroll_and_record |
Scroll through page, record video, capture screenshots |
capture_current_state |
Take screenshot at current scroll position |
Animation Analysis
| Tool | Description |
|---|---|
extract_animation_code |
Extract GSAP, ScrollTrigger, Lenis code (AST-based) |
deep_script_analysis |
Deep analysis of all JS for animation patterns |
get_scroll_triggers |
Get all ScrollTrigger instances and configs |
analyze_lenis |
Analyze Lenis smooth scroll configuration |
get_animation_css |
Extract CSS animations and keyframes |
Element Inspection
| Tool | Description |
|---|---|
watch_element |
Watch element for style changes during scroll |
get_element_animations |
Get all animations affecting a specific element |
trace_animation_execution |
Capture GSAP calls in real-time during scroll |
Interaction
| Tool | Description |
|---|---|
interact_with_page |
Click, hover, scroll to, or type on elements |
Output
| Tool | Description |
|---|---|
generate_report |
Generate JSON, Markdown, or HTML report |
export_replicable_code |
Generate standalone code (vanilla/React/Vue) |
Usage Examples
Basic Analysis Workflow
1. open_page(url: "https://example.com")
2. scroll_and_record()
3. extract_animation_code()
4. get_scroll_triggers()
5. generate_report(format: "markdown")
6. close_browser()
Watch a Specific Element
1. open_page(url: "https://example.com")
2. watch_element(selector: ".hero-title", scrollThrough: true)
3. get_element_animations(selector: ".hero-title")
Extract and Replicate
1. open_page(url: "https://example.com")
2. scroll_and_record()
3. extract_animation_code()
4. export_replicable_code(framework: "react")
Architecture
src/
├── index.ts # MCP server entry point
├── cli.ts # CLI entry point
├── types.ts # TypeScript interfaces
├── schemas.ts # Zod validation schemas
├── extractors/
│ ├── gsap-extractor.ts # AST-based GSAP extraction
│ ├── lenis-extractor.ts # Lenis extraction
│ └── css-extractor.ts # CSS animations (CSSOM)
├── browser/
│ ├── browser-manager.ts # Playwright lifecycle
│ └── page-controller.ts # Scroll, interact, capture
├── analyzers/
│ ├── runtime-analyzer.ts # GSAP/ScrollTrigger API queries
│ └── script-analyzer.ts # Orchestrates extractors
└── reporters/
├── markdown-reporter.ts
├── html-reporter.ts
└── code-exporter.ts
Development
# Build
npm run build
# Watch mode
npm run dev
# Run tests
npm test
# Test coverage
npm run test:coverage
How It Differs from GSAP Generation MCPs
| Feature | Generation MCPs | This Analyzer MCP |
|---|---|---|
| Purpose | Create new animations | Analyze existing sites |
| Input | Natural language intent | URLs/existing pages |
| Output | GSAP code to use | Analysis reports, extracted code |
| Browser | Not needed | Playwright for real rendering |
| Use case | Building new features | Learning, debugging, reverse engineering |
Technical Details
- Built with TypeScript and MCP SDK
- Uses Playwright for real browser automation
- AST parsing with Acorn (not regex)
- Zod for input validation
- Supports headless Chromium
- Records video automatically
- Injects tracers to capture runtime GSAP calls
Known Limitations
AST Extraction on Minified Sites
The extract_animation_code tool uses AST parsing which skips minified JavaScript by default. This is intentional because:
- Minified code is slow to parse and often fails
- Variable names are mangled, making extraction less useful
Workaround: Use get_scroll_triggers for runtime detection - it queries the GSAP/ScrollTrigger APIs directly and works regardless of code minification.
watch_element Style Changes
The watch_element tool may report fewer style changes than expected because browser scroll event listeners don't always fire during programmatic scrolls (window.scrollTo()). For comprehensive element tracking, combine with scroll_and_record which captures animation state at each frame.
Tested Sites
| Site | ScrollTriggers Found | Notes |
|---|---|---|
| gsap.com | 35 | Runtime detection works; AST returns 0 (minified) |
| Custom unminified sites | Full extraction | AST + runtime both work |
License
MIT
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
Neon Database
MCP server for interacting with Neon Management API and databases
E2B
Using MCP to run code via e2b.
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.