# tapsite [Health: Active]

**Category:** 💻 Developer Tools  
**Repository:** https://github.com/mgriffen/tapsite  
**GitHub Stars:** 3  
**npm Downloads (last month):** 204  
**Views:** 0  
**Installs:** 0  
**Upvotes:** 0  
**Directory Page:** https://allmcps.com/mcp/tapsite

## Description
MCP server for web inspection, design system extraction, and accessibility audits

## Tools
Capabilities this server exposes over MCP:

- **tapsite_login_manual** — Open headed browser for manual login (MFA). Call login_check when done.
- **tapsite_login_check** — Verify auth state after manual login. Returns title, URL, content preview.
- **tapsite_inspect** — Navigate to URL and/or inspect page DOM with numbered interactive elements. Optional screenshot.
- **tapsite_screenshot** — Screenshot the page. Optional element highlighting with numbered badges.
- **tapsite_interact** — Click, fill, select, check, or hover an indexed element. Returns updated DOM.
- **tapsite_scroll** — Scroll page or scroll an element into view.
- **tapsite_run_js** — Run JS in page context. Large results (>2KB) write to disk.
- **tapsite_close** — Close the browser session.
- **tapsite_extract_table** — Extract table data as structured rows.
- **tapsite_extract_links** — Extract all links with text and href.
- **tapsite_extract_colors** — Extract color palette sorted by frequency.
- **tapsite_extract_fonts** — Extract fonts: families, sizes, weights, sources.
- **tapsite_extract_css_vars** — Extract CSS custom properties, categorized by type.
- **tapsite_extract_spacing** — Extract spacing scale: margins, padding, gaps, radii.
- **tapsite_extract_images** — Extract all images with src, dimensions, alt, format.
- **tapsite_download_images** — Download images to disk. Uses session cookies for auth assets.
- **tapsite_extract_svgs** — Extract SVGs: inline markup, external URLs, icon/illustration classification.
- **tapsite_extract_favicon** — Extract favicon and icon references. Optional download.
- **tapsite_extract_layout** — Extract layout tree: flex/grid/block containers with properties.
- **tapsite_extract_components** — Detect repeated UI component patterns with instance counts.
- **tapsite_extract_breakpoints** — Extract CSS breakpoints and detect framework (Tailwind/Bootstrap/MUI).
- **tapsite_extract_metadata** — Extract metadata: OG, Twitter Cards, JSON-LD, RSS, canonical.
- **tapsite_extract_content** — Extract main content as clean markdown, stripping chrome.
- **tapsite_extract_forms** — Extract forms: fields, validation, actions, hidden fields.
- **tapsite_extract_animations** — Extract CSS animations, transitions, and detect animation libraries.
- **tapsite_extract_a11y** — Accessibility audit with score (0-100) and issues by severity.
- **tapsite_extract_darkmode** — Detect dark mode support. Optionally capture dark palette.
- **tapsite_extract_perf** — Performance metrics: Web Vitals, resource sizes, timing.
- **tapsite_extract_shadows** — Extract box-shadow and text-shadow patterns as design tokens.
- **tapsite_extract_icons** — Detect icon font libraries and extract icon usage (Font Awesome, Material, Bootstrap Icons, etc.).
- **tapsite_extract_contrast** — WCAG contrast ratio audit between text and background color pairs.
- **tapsite_extract_web_components** — Inventory custom elements, shadow DOM, exposed attributes, and web component libraries.
- **tapsite_extract_third_party** — Classify all cross-origin scripts and resources by vendor and category (analytics, payments, chat, etc.).
- **tapsite_extract_storage** — Audit client-side storage: cookies (with classification), localStorage, sessionStorage, IndexedDB.
- **tapsite_extract_pwa** — Profile PWA readiness: manifest, service worker, Apple/MS meta tags, and platform capabilities.
- **tapsite_extract_security** — Security audit: CSP, SRI, mixed content, iframe sandboxing, insecure forms, response headers, and scoring.
- **tapsite_extract_aiml** — Detect client-side AI/ML libraries (TensorFlow.js, ONNX, MediaPipe, etc.) and browser ML capabilities.
- **tapsite_extract_canvas** — Inventory canvas elements, detect 2D/3D frameworks (Three.js, PixiJS, Phaser, etc.), and report GPU info.
- **tapsite_extract_i18n** — Profile internationalization: languages, hreflang tags, RTL content, i18n frameworks, and language switchers.
- **tapsite_extract_graphql** — Detect GraphQL clients (Apollo, Relay, urql), related scripts, and endpoint hints.
- **tapsite_extract_wasm** — Detect WebAssembly modules, infer source languages (Rust, C++, Go, .NET), and check WASM capabilities.
- **tapsite_extract_markdown** — Extract page content as clean Markdown. Modes: raw (full page), fit (noise filtered), citations (numbered refs). Optional BM25 query filtering and chunking.
- **tapsite_extract_custom** — Extract structured data using a custom schema (CSS selectors, XPath, or regex patterns). Define your own extraction shape.
- **tapsite_extract_schema_suggest** — Analyze page DOM for repeated patterns and suggest a CSS extraction schema. Returns a schema you can refine and pass to tapsite_extract_custom.
- **tapsite_capture_network** — Capture network traffic for a duration. Filters static assets by default.
- **tapsite_extract_api_schema** — Infer API schemas from captured network traffic.
- **tapsite_extract_stack** — Detect tech stack: frameworks, libraries, CMS, analytics, CDNs.
- **tapsite_crawl** — BFS crawl with extraction per page. Writes to output/crawl-{ts}/.
- **tapsite_diff_pages** — Compare two URLs (cross-site) or the same URL over time (temporal). Runs real extractors — colors, fonts, spacing, a11y, perf, metadata, components, breakpoints — and produces a structured diff with regressions/improvements highlighted. Omit url2 for temporal mode (compares against last saved snapshot).
- **tapsite_export** — Export URLs as Markdown + HTML report + JSON + CSV + screenshots.
- **tapsite_export_design_report** — Full design system report: HTML + W3C tokens JSON + CSS vars.
- **tapsite_teardown** — Competitive design teardown — runs colors, fonts, CSS vars, spacing, shadows, components, breakpoints, animations, icons, stack detection, web components, third-party scripts, AI/ML, perf, a11y, contrast, and dark mode in one call.
- **tapsite_audit** — Pre-launch quality audit with scorecard — runs a11y, contrast, perf, metadata (SEO), dark mode, forms, and security. Returns pass/fail per category with an overall score out of 100.
- **tapsite_harvest** — Migration asset inventory — crawls up to maxPages pages and extracts content, images, SVGs, forms, fonts, and links per page. Returns per-page counts and a link map. Output written to output/harvest-{ts}/.
- **tapsite_designsystem** — Design system extraction — runs colors, fonts, spacing, shadows, CSS vars, breakpoints, animations, icons, and components. Exports W3C design-tokens.json, design-tokens.css, and raw-data.json to output/design-system-{ts}/.

## Claude Desktop Quick Installation
Install path detected from listing signals. Uses `npx` (confidence: high):

```json
"mcpServers": {
  "tapsite": {
    "command": "npx",
    "args": ["-y","tapsite"]
  }
}
```

## Documentation

## What the tapsite MCP server does

The tapsite MCP server gives AI agents structured inspection tools for websites rather than limiting them to browser actions. It can inspect page structure, extract readable content and metadata, identify links, forms, tables, images, SVGs, favicons, and convert pages to Markdown. Custom extraction supports CSS selectors, XPath, regular expressions, and suggested schemas based on the page DOM.

Its design-system features collect colors, fonts, spacing values, CSS custom properties, shadows, layout information, repeated components, responsive breakpoints, and icon usage. A design report can be exported as a visual HTML report, W3C design-token JSON, and CSS variables.

Quality and discovery tools cover accessibility and contrast checks, Core Web Vitals and resource sizes, security headers, dark mode, PWA signals, internationalization, frontend frameworks, third-party services, client-side storage, GraphQL, AI/ML integrations, canvas and WebGL, WebAssembly, web components, animations, and observed API schemas. Crawling, cross-site comparison, and temporal snapshots support multi-page or change-tracking work.

## How it works

The server runs a browser-backed extraction engine with Playwright. Session tools can open a headed browser for manual login, check authentication, inspect and interact with indexed elements, scroll, run JavaScript, take screenshots, and close the session. Login state persists across tool calls, so protected pages can be inspected after a user completes authentication.

Page extraction can run in parallel through browser pooling. Disk caching allows crawls to resume and lets results be reused across sessions. The engine also documents a three-level anti-bot escalation process with automatic proxy rotation for protected sites, although access to a particular website will depend on that site’s behavior.

## Setup and configuration

Run the server without a local installation using `npx -y tapsite`, or install it globally with `npm install -g tapsite`. The documented MCP configuration starts it with the `npx` command and the `-y tapsite` arguments.

Playwright may download Chromium automatically on the first run, with an approximately 150 MB one-time download. No required environment variables are documented.

## Tools and capabilities

The tapsite MCP server exposes 55 tools across session control, content and custom extraction, design tokens, visual assets, layout, quality checks, technology detection, network capture, multi-page analysis, and export. Four higher-level workflows combine these capabilities: a competitive design teardown, a pre-launch audit, a site asset inventory, and design-system extraction.

Exports include JSON, Markdown, HTML reports, CSV tables, screenshots, W3C design tokens, and CSS variables. The server is configured as a standard stdio MCP process and is documented for Claude, Cursor, Windsurf, and other MCP-compatible agents.

## Limitations and notes

Credentials are intended to be entered through the manual browser login flow rather than sent through chat. LangChain and LangGraph support is described as upcoming, not as an available integration.

_Full upstream README: https://allmcps.com/mcp/tapsite/readme_