Bocha Search MCP Server.
Configuration
View docs{
"mcpServers": {
"bochaai-bocha-search-mcp": {
"command": "uv",
"args": [
"--directory",
"/path/to/bocha-search-mcp",
"run",
"bocha-search-mcp"
],
"env": {
"BOCHA_API_KEY": "YOUR_API_KEY"
}
}
}
}Bocha MCP Server enables your AI applications to query Bocha’s search services and receive web results plus structured modal cards for domain-rich content. It’s designed to be used by MCP clients to fetch high-quality information from Bocha’s extensive content sources and present it in a usable, semantically rich format.
You integrate the Bocha MCP Server with an MCP client to perform searches and retrieve both standard web results and structured modal cards (such as weather,百科, medical, and more). Start the local MCP server, point your MCP client at it, and then issue search requests from your AI workflow. The server handles querying Bocha’s search API behind the scenes and returns results ready for your AI system to consume.
Prerequisites: ensure you have a modern runtime environment and Git installed.
# Step 1: Clone the Bocha MCP repository
git clone [email protected]:BochaAI/bocha-search-mcp.git
# Step 2: Install dependencies (if applicable for your setup)
# This step may vary by environment (e.g., npm install, yarn install, or others)Configuration is focused on running a local MCP server using the provided runtime command. The server is intended to be launched from your MCP client environment with the appropriate directory context.
# Step 1 example: standard runtime configuration for Claude Desktop
# This snippet demonstrates how to reference the local Bocha MCP server
"mcpServers": {
"bocha-search-mcp": {
"command": "uv",
"args": [
"--directory",
"/path/to/bocha-search-mcp",
"run",
"bocha-search-mcp"
],
"env": {
"BOCHA_API_KEY": "sk-****"
}
}
}
```
```
# Step 2: Optional local debugging command
npx @modelcontextprotocol/inspector uv --directory /path/to/bocha-search-mcp run bocha-search-mcpQuery Bocha Web Search to retrieve titles, URLs, summaries, site names, icons, and publish times for web results.
Extend web search results with AI-driven semantic understanding and generate structured modal cards for vertical content like weather, calendars, and encyclopedias.