crawlora-mcp

Crawlora MCP is a hosted Model Context Protocol (MCP) server designed to provide structured public web data to AI clients and agents. It offers 3631 tools across 471 platform groups, covering categories such as search, maps, e-commerce, social, finance, travel, app stores, media, and reviews, with each tool returning clean, normalized JSON rather than HTML that needs parsing.

Key capabilities

• Provides 3631 structured data tools, covering multiple platform groups (such as 1stdibs, 7now, abcau, abcnews, adidas, agoda, airbnb, amazon, apple-maps, appstore, and more).
• All tools return normalized JSON fields, making them easy for agents to use directly without writing parsers.
• Supports any MCP-compatible client, including Claude, Cursor, VS Code (GitHub Copilot), Windsurf, Gemini, Cline / Roo Code, and more.

How to use

There are two ways to use it:

1. Hosted endpoint (recommended): Connect any MCP client to the hosted endpoint https://mcp.crawlora.net/mcp(Streamable HTTP), using CRAWLORA_API_KEY for authentication.
2. Run locally (open-source server): Clone the repository, run the local stdio MCP server (index.mjs), start it via npx or Docker, with each tool wrapping the Crawlora REST API (https://api.crawlora.net/api/v1), also using CRAWLORA_API_KEY for authentication.

Client configuration examples

• Claude Code (CLI): After adding, it takes effect in the local scope; you can use --scope user or --scope project to adjust the scope, and confirm with the command claude mcp list.
• Claude Desktop: In Settings → Developer → Edit Config, add to claude_desktop_config.json.
• Cursor: Project-local .cursor/mcp.json or global ~/.cursor/mcp.json.
• VS Code (GitHub Copilot): Project .vscode/mcp.json or user mcp.json。
• Windsurf：~/.codeium/windsurf/mcp_config.json(note the serverUrl key).
• Gemini:~/.gemini/settings.json(note the httpUrl key).
• Cline / Roo Code: Add to cline_mcp_settings.json, using the generic remote configuration.
• Other Streamable HTTP clients: Use the generic configuration in examples/mcp.json .
• stdio-only clients (such as Zed): Run the local server and point to its npx command.

Local server configuration (stdio)

Tool definitions are in tools.json , generated from the API catalog published by Crawlora; each tool includes a real input schema and maps to a single REST endpoint.

Authentication

Authenticate using a Crawlora API key. A missing or invalid API key will return a 401 error. It is recommended to store the key in an environment variable to avoid pasting or committing it directly.

Use cases

• When AI agents need to obtain structured web data (such as search, e-commerce, social media, or financial data).
• When developers want to integrate multiple data sources into a single MCP endpoint, avoiding stitching together multiple APIs (such as SerpApi, Firecrawl, ScraperAPI, or ScrapingBee).
• When AI clients need to call data tools directly without writing additional parsing logic.

Limitations

• Usage requires a Crawlora API key; free accounts receive 2000 credits per month, with no credit card required.
• The tool count in the repository's current source directory may differ from the count in the npm package crawlora-mcp until a tagged version is released.
• The service itself is a hosted SaaS and is subject to Crawlora's terms; the repository only contains connection documentation and a manifest.
• For more information such as specific pricing and credit consumption rates, please refer to the pricing page.

Resources

• Website:https://crawlora.net/mcp
• Documentation and Playground:https://crawlora.net/docs · https://crawlora.net/playground
• Pricing:https://crawlora.net/pricing
• Server card:https://crawlora.net/.well-known/mcp/server-card.json
• Official MCP registry listing:server.json
• License: see LICENSE