SearchAPI is a lightweight, self-hostable REST API that wraps DuckDuckGo’s meta-search engine (DDGS) and the Crossref academic index to deliver structured search results over HTTP. It was built to solve a specific problem: collecting large volumes of relevant URLs across different content types (web pages, PDFs, research papers, books, news articles, GitHub repositories) for building LLM fine-tuning datasets — without hitting paywalls or rate limits from proprietary search APIs.Documentation Index
Fetch the complete documentation index at: https://mintlify.com/pratyay360/searchapi/llms.txt
Use this file to discover all available pages before exploring further.
Quickstart
Make your first API call in under 5 minutes
API Reference
Browse all endpoints with parameters and examples
MCP Server
Connect SearchAPI to Claude, Cursor, and other AI assistants
Dataset Generation
Strategies for building LLM training datasets
What SearchAPI provides
SearchAPI exposes a set of focused endpoints, each targeting a different content type. All endpoints return arrays of URLs (or DOIs for papers) as JSON, making them easy to integrate into data pipelines, scripts, or AI agent workflows.Web search
General-purpose meta-search across multiple engines
Engine-specific search
Target Bing, Google, DuckDuckGo, Brave, Yandex, and more
Research papers
Academic paper search via Crossref, returns DOIs
Books
Book search returning direct URLs
News
Current news articles by topic
PDFs & filetypes
Search for PDFs, DOCX, PPTX, and any filetype
Repositories
Find GitHub and GitLab repositories
Wikipedia & Wikimedia
Aggregated results from Wikipedia and related Wikimedia sites
Key characteristics
- No API key required — SearchAPI uses DuckDuckGo’s DDGS library under the hood. No account, no billing, no quota management on your end.
- MCP server built in — The same endpoints are also available as MCP tools, letting AI assistants like Claude call them directly.
- Docker and Podman ready — A single
podman-compose up -dstarts the service. A prebuilt Docker image is available on Docker Hub. - One-click Render deploy — The included
render.yamlconfigures automatic deployment on Render’s free tier. - CORS open by default — All origins, methods, and headers are permitted out of the box, so you can call the API from any frontend or script.
Getting started
Deploy or run locally
Pull and start the container, or clone the repo and run with
uv fastapi run. See the deployment guide for all options.Make your first search call
Call
GET /search/?query=machine+learning&limit=10 to get a list of URLs. See the quickstart for a full walkthrough.Automate your data collection
Use the endpoint that matches your content type — papers, PDFs, news, books, or repositories — and pipe the results into your dataset pipeline.
Optionally connect to an AI assistant
Mount the MCP server and give Claude or Cursor direct access to all search tools. See the MCP server guide.
SearchAPI is intended for educational and research use. You are responsible for complying with the terms of service of any upstream search engine and applicable laws in your jurisdiction.