Litescrape MCP Server is a server for MCP (Model Context Protocol) clients that provides the results of multiple search and map services to AI clients in structured JSON form.
Purpose
The server enables any MCP client to obtain query results from Google Search, Bing, DuckDuckGo, and Google Maps. It can be used free of charge without an API key. The results come directly from the search engines' own pages and are returned by the Litescrape API as structured JSON. Links are real destination addresses, not redirects; nothing is cached, and each call is fetched in real time.
Main capabilities
- Usable without a key: Google Search, Bing, DuckDuckGo, Google Maps results.
- Unlocked after adding an API key: google_ai_mode, google_ai_overview, google_shopping, google_reviews, and web_fetch; removes the daily limit and allows concurrent calls.
- Each call can pass the result_groups parameter to specify which top-level result groups to keep (for example ["organic_results", "knowledge_graph"]), making it easy to control the model context size; search_metadata is always included.
- The tool result is a single text block: a one-line summary, a blank line, and then the API's JSON response returned as-is. The same JSON is also returned as structuredContent for clients that read that field.
Usage
- Requires Node.js 20 or later.
- No account and no key required; a one-line configuration is enough for Cursor, Windsurf, Claude Desktop, Codex, Gemini CLI, and most other clients (mcp.json, claude_desktop_config.json, or the client's MCP settings).
- Hosted endpoint: clients that support connecting to a remote server can use https://mcp.litescrape.com/mcp，工具与免费额度相同；发送 Authorization: Bearer ls_live_... unlocks the remaining features.
- Add with Claude Code: claude mcp add litescrape -e LITESCRAPE_API_KEY=ls_live_... -- npx -y litescrape-mcp-server
- Add with the hosted endpoint: claude mcp add --transport http litescrape https://mcp.litescrape.com/mcp --header "Authorization: Bearer ls_live_..."
- Parameter names and accepted values follow the Litescrape API reference documentation.
- Streaming HTTP mode: litescrape-mcp-server --http (or LITESCRAPE_MCP_TRANSPORT=http) serves MCP over HTTP instead of stdio: stateless POST /mcp, GET /healthz, listens on PORT (default 8080) and HOST (default 0.0.0.0), or use --port and --host. Each request's API key comes from its Authorization: Bearer header or the ?api_key= query parameter, so one process can serve both keyed and keyless callers at the same time; in this mode LITESCRAPE_API_KEY is ignored. This is exactly https://mcp.litescrape.com/mcp The mode in which it runs.
- Shared address scenario: keyless callers are collectively counted toward one quota. LITESCRAPE_KEYLESS_PROXY_SECRET is the secret shared by the Litescrape API and its own hosted endpoint, used to forward each caller's address; LITESCRAPE_EDGE_SECRET is the value added by Cloudflare in front of that endpoint so that the caller can be read from CF-Connecting-IP. Private deployments do not need these two, and in that case the server's own address is metered.
- Each request identifies itself via X-Litescrape-Client: mcp/<version> and the litescrape-mcp-server/<version> user agent. The server writes protocol messages to stdout and logs to stderr.
- Development: npm run dev runs the server from source with tsx. Releases are tagged with the package version (for example, 0.1.0); the release workflow builds and publishes to npm with provenance.
Free quota and limits
- Without a key, only one call runs at a time per network, and failed calls are not counted. When the limit is reached, the tool result explains this and tells the model how to add a key, making it easy for the agent to relay to the user.
- Requests time out after 120 seconds by default (the API's own deadline is 90 seconds). AI Mode is the slowest tool.
Applicable scenarios
Suitable for scenarios where an AI client needs real-time structured data from search engines and maps, including search, map queries, shopping and review information, as well as using AI Mode, AI Overviews, and web scraping via a key.
Limitations
No features, charges, pricing, or other platform support information beyond the above is provided.
