OpenClaw¶

OpenClaw is a self-hosted personal AI assistant that connects to WhatsApp, Telegram, Discord, Slack, Signal, iMessage, and more. It uses skills — markdown-based instruction files — to teach the agent how to use tools.

Open Agent Search integrates with OpenClaw as a skill, giving your lobster assistant the ability to search the web, images, videos, news, books, and extract content from URLs.

Method 1: Skill File (Recommended)¶

OpenClaw skills live in your workspace. Create a skill folder and SKILL.md file:

1. Create the skill directory:

mkdir -p ~/.openclaw/workspace/skills/open-agent-search

2. Create ~/.openclaw/workspace/skills/open-agent-search/SKILL.md:

---
name: open-agent-search
description: Search the web, images, videos, news, and books via Open Agent Search MCP server. Privacy-first, powered by DuckDuckGo.
metadata: {"openclaw": {"requires": {"bins": ["uvx"]}, "emoji": "🔍"}}
---

# Open Agent Search

You have access to a powerful search MCP server called Open Agent Search.

## Starting the server

Run the MCP server via stdio:

```bash
uvx --from open-agent-search oas-mcp

Available tools¶

• search_web — Text / web search
• search_images — Image search with size & colour filters
• search_videos — Video search with resolution & duration filters
• search_news — News search with time-limit filter
• search_books — Book search
• search_everything — Parallel search across all sources
• fetch_content — Extract content from a single URL
• fetch_multiple_contents — Extract content from multiple URLs (max 10)

Usage¶

When the user asks you to search for something, use these tools. Prefer search_web for general queries and search_everything for comprehensive research.

**3.** Restart your OpenClaw gateway or start a new session. The skill will be picked up automatically.

!!! tip "Prerequisites"
    You need [`uv`](https://docs.astral.sh/uv/getting-started/installation/) installed. The `uvx` command comes with it.

---

## Method 2: MCP Server in `openclaw.json`

You can also wire Open Agent Search as an MCP tool server directly in your OpenClaw configuration.

**1.** Edit `~/.openclaw/openclaw.json`:

```json
{
  "tools": {
    "mcp": {
      "open-agent-search": {
        "command": "uvx",
        "args": ["--from", "open-agent-search", "oas-mcp"],
        "transport": "stdio"
      }
    }
  }
}

2. Restart the OpenClaw gateway:

openclaw gateway --port 18789

All 8 search tools will be available to your OpenClaw assistant.

Method 3: Remote HTTP Server¶

If you have a deployed Open Agent Search instance (e.g. on Vercel or Docker), point OpenClaw at the HTTP MCP endpoint:

{
  "tools": {
    "mcp": {
      "open-agent-search": {
        "url": "https://your-instance.vercel.app/ai/mcp",
        "transport": "http"
      }
    }
  }
}

This is useful when your OpenClaw runs on a different machine from your search server, or when you want to share a single deployed instance.

Install via ClawHub¶

If the skill is published on ClawHub, you can install it directly:

clawhub install open-agent-search

This downloads the skill into your workspace automatically.

Verify¶

Once configured, talk to your OpenClaw assistant:

"Search the web for the latest news about open source AI agents"

Your assistant should use the search_web or search_news tool from Open Agent Search and return results.

You can also check the OpenClaw Control UI at http://127.0.0.1:18789 to see the skill listed under your active skills.

Available Tools¶

Once connected, the server exposes 8 tools. See the Tools Reference for full parameter documentation.