generated: '2026-09-19'
method: searched
status: published
source: https://webcrawlerapi.com/docs/sdk/mcp + https://github.com/WebCrawlerAPI/webcrawlerapi-mcp (src/index.ts
  read 2026-09-19) + https://registry.npmjs.org/webcrawler-mcp
summary: 'WebCrawlerAPI ships an official MCP server as an npm package (webcrawler-mcp) that a user runs locally
  over stdio, or self-hosts over HTTP with USE_HTTP=true. There is NO provider-hosted remote endpoint: the docs
  give no URL, mcp.webcrawlerapi.com does not resolve, and the HTTP mode in the source binds a local port (default
  8080) on the user''s machine. The server exposes exactly one tool, webcrawler-scrape, which POSTs to https://api.webcrawlerapi.com/v2/scrape
  with the user''s API key from the WEBCRAWLER_API_KEY environment variable.'
deployment:
  mode: local-stdio
  endpoint: null
  install: npx -y webcrawler-mcp
  package: https://www.npmjs.com/package/webcrawler-mcp
  repository: https://github.com/WebCrawlerAPI/webcrawlerapi-mcp
  auth: api-key
  verified: searched
  verified_note: Established from the provider's docs page, the npm registry metadata (version 1.0.0, published
    2025-07-19, maintainer niiotyo - the same maintainer as webcrawlerapi-js and n8n-nodes-webcrawlerapi) and the
    TypeScript source. Not executed by this pipeline; no remote endpoint exists to probe.
  self_hosted_http:
    env:
      USE_HTTP: 'true'
      PORT: 8080 (default)
    paths:
    - /mcp
    - /
    - /health
    note: Streamable-HTTP mode served by the same package on the user's own host. This is still not an agent surface
      until someone runs it - it is recorded so the "both" mode is not mistakenly claimed.
server:
  name: webcrawler-mcp
  transport: stdio (default) | streamable-http (self-hosted)
  sdk: '@modelcontextprotocol/sdk ^1.0.0'
  language: TypeScript
  client_config_example:
    mcpServers:
      webcrawler:
        command: npx
        args:
        - -y
        - webcrawler-mcp
        env:
          WEBCRAWLER_API_KEY: <your API key>
auth:
  methods:
  - api-key
  apikey:
    env: WEBCRAWLER_API_KEY
    upstream_header: 'Authorization: Bearer {api_key}'
    obtain: https://dash.webcrawlerapi.com/access
tool_count: 1
tools:
- name: webcrawler-scrape
  category: scrape
  description: Scrape content from a webpage using WebcrawlerAPI. Uses WEBCRAWLER_API_KEY environment variable for
    authentication.
  inputSchema:
    type: object
    properties:
      url:
        type: string
        description: The URL of the webpage to scrape
      prompt:
        type: string
        description: Optional prompt to extract specific information from the page
      main_content_only:
        type: boolean
        description: 'Whether to extract only the main content of the page (default: false)'
    required:
    - url
  inputSchema_source: zod shape in src/index.ts (z.string / z.string().optional() / z.boolean().optional())
  backing_rest: POST https://api.webcrawlerapi.com/v2/scrape
  returns: markdown text plus page title and HTTP status; structured_data when a prompt is given
coverage_gap: The MCP server covers one of 24 REST operations. Multi-page crawl jobs, feeds, the crawling agent
  and organization usage are REST/SDK/CLI-only - see mcp/webcrawlerapi-com-tool-crosswalk.yml. The provider also
  publishes a Claude Code Agent Skill (skills/) that reaches the crawl surface via curl, which is a different agent
  surface from MCP.
currency:
  npm_version: 1.0.0
  npm_published: '2025-07-19'
  repo_last_push: '2025-09-11'
  note: The GitHub repo moved after the only npm release (HTTP-mode tests were added), so npx users run a build
    that predates the repo head.
listings:
  docs: https://webcrawlerapi.com/docs/sdk/mcp
  changelog: https://webcrawlerapi.com/changelog/2025-07-27-mcp-integration