open-agent-search 1.0.3

A production-ready metasearch API and MCP server powered by DuckDuckGo

Open Agent Search (OAS)

The sensory organ for the AI agent ecosystem. An independent, privacy-first search layer providing unfiltered access to the world's knowledge via MCP and HTTP API.

Open Agent Search is an independent foundational tool. We are not bound to any single agent framework but support them all (including OpenClaw via openclaw-search). Our mission is accuracy, speed, and universal access.

Read the full documentation →

---

Quick Start

Zero-install (MCP)

# Works instantly — no clone or install needed
uvx --from open-agent-search oas-mcp

Install & Run

# Install from PyPI
uv pip install open-agent-search

# Start the HTTP API + MCP server
open-agent-search
# → http://localhost:8000 (API)
# → http://localhost:8000/ai/mcp (MCP)

From Source

git clone https://github.com/Jayanth-MKV/open-agent-search.git
cd open-agent-search
uv sync
uv run open-agent-search

Docker

docker compose up -d

---

What's Included

Feature	Description
REST API	8 search endpoints (text, images, videos, news, books, unified, content fetch)
MCP Server	8 tools accessible via stdio or HTTP — works with every major AI coding client
OpenClaw Compatible	Works as an OpenClaw skill via stdio MCP transport
Rate Limiting	Per-IP rate limits with configurable dev/prod profiles
One-Click Deploy	Deploy to Vercel in seconds
Docker	Multi-stage Dockerfile + Compose with health checks

---

MCP Client Setup

Full setup guides →

Add to any MCP client with uvx — zero install:

{
  "mcpServers": {
    "open-agent-search": {
      "command": "uvx",
      "args": ["--from", "open-agent-search", "oas-mcp"]
    }
  }
}

Guides available for: Claude Desktop · Claude Code · Cursor · VS Code · Windsurf · OpenClaw · and more

---

API Endpoints

Endpoint	Method	Description
/api/search/text	GET	Web / text search
/api/search/images	GET	Image search
/api/search/videos	GET	Video search
/api/search/news	GET	News search
/api/search/books	GET	Book search
/api/search/all	GET	Unified parallel search
/api/content/fetch	GET	Fetch & extract content from a URL
/api/content/fetch-multiple	POST	Fetch content from multiple URLs
/ai/mcp	—	MCP server endpoint

Full API reference →

---

Documentation

• Installation
• Quick Start
• MCP Integration
• API Reference
• Docker Deployment
• Vercel Deployment
• Contributing

Contributing

See CONTRIBUTING.md.

License

MIT