55 Tools.
One MCP Server.
Total Web Intelligence.
Tap into any website's DNA — design tokens, accessibility scores, tech stacks, security headers — all returned as structured JSON your AI agent can act on. Your agent decides which tools to call, combines the results, and presents them however you ask — a competitive teardown deck, an accessibility audit report, a design token file ready to drop into your codebase. 55 extraction tools with custom CSS, XPath, and regex selectors. Anti-bot resilience with automatic proxy rotation. Persistent caching across sessions. Runs locally. Supports authenticated sessions behind MFA and SSO. Zero config.
Mar 24, 2026 · 14:32 UTC
3.2s · 55 tools executed
Key Capabilities
Design System Extraction
Extract color palettes, typography rules, spacing scales, CSS variables, shadows, and component patterns from any website. Get the complete design DNA as structured JSON.
Authenticated Browsing
Log in once through a headed browser — MFA, SSO, whatever the site requires. Session persists across all tool calls. Credentials never touch the wire.
Prompt Injection Defense
Hidden element filtering at extraction time plus output sanitization before anything reaches your LLM. Two layers, zero trust.
Parallel Browser Pooling
Multiple browser contexts for concurrent page extraction. Crawl entire sites faster with managed pool slots.
Persistent Disk Caching
Crawl results cached with TTL. Interrupted crawls resume automatically. Results reused across sessions.
Anti-Bot Escalation
3-tier system: default browser, stealth scripts, isolated contexts with proxy rotation. Automatic per-domain escalation.
Security & Accessibility Audits
CSP scoring, SRI checks, WCAG compliance, contrast ratios — actionable letter grades, not just raw data.
Tech Stack Intelligence
Framework detection, third-party script classification, storage profiling, WASM analysis — know what's under the hood.
55 Tools, 11 Categories
Every tool returns structured JSON designed for LLM context windows. One MCP server, modular by design.
Session
Navigate, authenticate, interact, screenshot — persistent browser context across calls.
Content Extraction
Tables, links, metadata, forms, and clean Markdown with relevance filtering and chunking.
Custom Extraction
Bring your own CSS, XPath, or regex selectors — plus auto-schema suggestion from page structure.
Design Tokens
Colors, fonts, CSS variables, spacing scales, and shadow patterns as structured JSON or W3C tokens.
Visual Assets
Images, SVGs, favicons — inventory with dimensions, formats, and bulk download.
Layout Intelligence
Layout trees, component detection, responsive breakpoints and framework detection.
Quality & Compliance
Accessibility, contrast, performance, security headers, dark mode, PWA readiness, i18n.
Technology Detection
Frameworks, animations, icons, GraphQL, AI/ML, canvas, WASM, web components, third-party scripts, storage.
Network & API
Capture traffic and infer API schemas from observed requests.
Multi-page
Crawl sites with parallel browser pooling and diff pages over time with cached snapshots.
Export
JSON, Markdown, HTML reports, CSV tables, screenshots, and W3C design tokens.
Workflows
One-command presets — full audit, harvest, design system, teardown.
Get Started in 60 Seconds
Three steps to web intelligence in your AI workflow.
1 Install
That's it. No API keys, no accounts, no config files.
2 Add to your MCP client
Works with Claude Code, Cursor, Windsurf, or any MCP-compatible client.
3 Ask your agent
Your agent calls the right tools and returns structured JSON. You decide what to do with it.
"Extract the design system from stripe.com"