Skip to main content
Glama

Crawlee MCP

Un servidor local del Protocolo de Contexto de Modelo para obtener, rastrear y capturar pantallas web con Crawlee. Se ejecuta sobre stdio, por lo que funciona con Claude Code, Claude Desktop, Codex, Cursor y cualquier cliente compatible con MCP.

Herramientas

Herramienta

Descripción

fetch_url

Obtiene una sola URL y devuelve markdown limpio

fetch_multiple

Obtiene múltiples URLs en paralelo

crawl

Rastrea un sitio, descubre enlaces y devuelve contenido de página

screenshot

Toma una captura de pantalla PNG de una página

ping

Verificación de estado

Related MCP server: weblens-mcp

Instalación

Requiere Node.js 20 o superior. El primer uso de representación JavaScript o capturas de pantalla también requiere el navegador Chromium:

npx playwright install chromium

Una vez publicado en npm, cada cliente puede iniciarlo con:

npx -y @patrikmichi/crawlee-mcp

Hasta que el paquete npm esté publicado, instálelo directamente desde GitHub:

npx -y github:patrikmichi/crawlee-mcp

Configuración del cliente

Claude Code

claude mcp add --transport stdio crawlee -- npx -y @patrikmichi/crawlee-mcp

Claude Desktop

Añada esto a su archivo de configuración de Claude Desktop:

{
  "mcpServers": {
    "crawlee": {
      "command": "npx",
      "args": ["-y", "@patrikmichi/crawlee-mcp"]
    }
  }
}

Codex

Añada esto a ~/.codex/config.toml:

[mcp_servers.crawlee]
command = "npx"
args = ["-y", "@patrikmichi/crawlee-mcp"]

Otros clientes MCP

Use el mismo comando y argumentos: npx -y @patrikmichi/crawlee-mcp.

Configuración

cp env.example .env

Variable

Requerido

Descripción

CRAWLEE_PROXY_URLS

No

URLs de proxy separadas por comas para rotación

CRAWLEE_TIMEOUT_MS

No

Tiempo de espera de solicitud en ms (predeterminado: 30000)

CRAWLEE_STEALTH

No

Establecer en false para deshabilitar el modo sigiloso (predeterminado: true)

Desarrollo

npm install
npm start          # Run with tsx (no build needed)
npm run build      # Compile TypeScript
npm run typecheck  # Type check only

Este servidor es intencionalmente local y solo stdio porque la representación JavaScript y las capturas de pantalla requieren un navegador Playwright local.

Available Tools

5 tools
crawlA

Crawl a website starting from a seed URL, following discovered links across multiple pages. Returns an array of {url, title, content} objects. Use link_pattern to restrict crawling to specific URL patterns (e.g. "https://example.com/blog/**").

Inputs: url (required), label, max_pages, render_js, +3 more.

fetch_multipleA

Fetch multiple URLs in parallel and return an array of {url, content, error} objects. Useful for bulk research or comparison tasks.

Inputs: urls (required), render_js.

fetch_urlB

Fetch a web page and return its content as clean markdown. Use render_js=true for JavaScript-heavy pages (SPAs, Twitter/X, etc). Use render_js=false for fast HTML-only fetching.

Inputs: url (required), render_js, extract_content.

pingB

Check that the crawlee MCP server is running.

Inputs: no named parameters.

screenshotB

Take a screenshot of a web page and return it as a base64-encoded PNG image.

Inputs: url (required), full_page.

Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections.

Loading tool schema changelog…

    TDQS

    A3.8/5.0

    Scored across 5 tools

    Disambiguation5/5

    Each tool has a clearly distinct purpose: fetching a single page, fetching multiple pages in parallel, taking a screenshot, health checking, and crawling a website. There is no overlap or ambiguity between the tools.

    Naming Consistency5/5

    All tool names follow a consistent verb-based pattern using lowercase and underscores where needed (e.g., fetch_url, fetch_multiple). No mixing of styles like camelCase or inconsistent verb forms.

    Tool Count5/5

    With 5 tools, the server is well-scoped for web crawling and scraping. Each tool earns its place: single fetch, batch fetch, screenshot, full crawl, and a health check. The count is neither too thin nor excessive.

    Completeness4/5

    The tool set covers the core workflows of fetching pages and crawling sites, with parallel fetching and screenshot support. Minor gaps exist, such as advanced extraction capabilities or session management, but the surface is practical for typical use cases.

    Related MCP Connectors

    Related MCP Servers

    • A
      license
      Not graded
      quality
      C
      maintenance
      Enables web scraping and crawling capabilities for LLM clients, supporting single-page scraping, multi-page website crawling, and web search with multiple engines (Playwright, Cheerio, Puppeteer) and flexible output formats including markdown, HTML, text, and screenshots.
      7
      MIT
    • A
      license
      A
      quality
      D
      maintenance
      Enables AI agents to fetch and render web pages (including JavaScript-heavy SPAs) with headless Chromium, extract readable content with Mozilla Readability, capture navigation links, download images, and return a clean markdown file path.
      1
      17 npm
      ISC