one-search

OneSearch MCP Server provides web search, local browser search, scraping, crawling, and extraction capabilities through a modular set of tools. The server integrates multiple search providers (including local browser-based search via agent-browser, and optional external providers such as searxng, duckduckgo, bing, tavily, google, zhipu, exa, and bocha). The enabled tools are one_search for performing unified web searches, one_scrape for scraping content, one_map for map-related data extraction, and one_extract for structured data extraction from multiple URLs. You can configure the provider via environment variables (SEARCH_PROVIDER, SEARCH_API_URL, SEARCH_API_KEY) to switch between local browser searches and external search engines. The server runs locally with Chromium-based browser automation, enabling private, API-free browsing and scraping where possible.

Prerequisites:

• A Node.js environment (for npm/npx usage) or a container runtime if you prefer Docker-based deployment.
• A Chromium-based browser installed locally, or rely on agent-browser to install one (as per the project prerequisites).
• Optional: Docker installed if you want to run via the Docker image.

Installation options:

Option A: Claude Code CLI (Recommended for Claude users)

# Add to Claude Code with default settings (local search)
claude mcp add one-search-mcp -- npx -y one-search-mcp

# Add with custom search provider (e.g., SearXNG)
claude mcp add one-search-mcp -e SEARCH_PROVIDER=searxng -e SEARCH_API_URL=http://127.0.0.1:8080 -- npx -y one-search-mcp

# Add with Tavily API
claude mcp add one-search-mcp -e SEARCH_PROVIDER=tavily -e SEARCH_API_KEY=your_api_key -- npx -y one-search-mcp

Option B: Manual installation

# Install globally (Optional)
npm install -g one-search-mcp

# Or run directly with npx
npx -y one-search-mcp

Option C: Using Docker

# Pull the image (GitHub Container Registry or Docker Hub)
docker pull ghcr.io/yokingma/one-search-mcp:latest
# or
docker pull zacma/one-search-mcp:latest

Configuration (example):

{
  "mcpServers": {
    "one-search-mcp": {
      "command": "docker",
      "args": ["run", "-i", "--rm", "ghcr.io/yokingma/one-search-mcp:latest"],
      "env": {
        "SEARCH_PROVIDER": "local"
      }
    }
  }
}

Environment variables to consider:

• SEARCH_PROVIDER: one of searxng, duckduckgo, bing, tavily, google, zhipu, exa, bocha, local. Default is local.
• SEARCH_API_URL: URL of the API for the chosen provider (e.g., SearXNG API URL or Google Custom Search Engine ID).
• SEARCH_API_KEY: API key required by some providers (tavily, bing, google, zhipu, exa, bocha).

Self-hosting SearXNG, if desired, is covered in the README with Docker-based deployment steps. Ensure a Chromium-based browser is available or install via agent-browser when required.

Tips:

• If you encounter a Browser not found error, install a Chromium-based browser or run npx agent-browser install to provision one automatically.
• The migration notes indicate that Firecrawl has been removed in favor of agent-browser; environment variables FIRECRAWL_API_URL and FIRECRAWL_API_KEY are no longer used.
• For local browser searches, the server will automatically detect browsers already installed (Chrome, Edge, Chromium, Chrome Canary).
• When using Docker, you can pass environment variables to configure the search provider without changing the container image.
• If you plan to use SearXNG, you can deploy it locally and point ONE-SEARCH to http://127.0.0.1:8080 via SEARCH_API_URL.
• Ensure your system has enough resources (RAM) when running browser-based tasks, as scraping and rendering can be memory-intensive.