MCP Services
Z Code integrates the MCP (Model Context Protocol) service, providing you with an Agent collaboration and management protocol framework. By standardizing Agent workflows, you can easily extend the boundaries of Agent capabilities, enabling them to perform advanced tasks such as visual recognition, internet search, and web reading.
- Built-in MCP: We provide preset MCP services covering various common scenarios. Simply enable them in settings to quickly apply best practices.
- Custom Configuration: You can also create and configure your own MCP services to meet unique project needs, achieving maximum control and flexibility.
Configuration Guide
Entry Point
Go to Z Code Settings and click MCP Servers in the left navigation bar to access the management interface.

Interface
In the configuration interface, you can view the list of all integrated services and fine-tune specific services through the JSON editor.

Core MCP Service Configuration
1. Visual Understanding MCP (zai-mcp-server)
Empowers the Agent to understand and analyze image content.
- Refer to the BigModel configuration documentation to get your API Token.
- In Z Code Settings -> MCP Servers, click edit for
zai-mcp-server. - Fill in your
Z_AI_API_KEYin the environment variables (env) and setZ_AI_MODEtoZHIPU. - Save and enable.

2. Internet Search MCP (web-search-prime)
Enables the Agent to access the internet in real-time for the latest information and data.
- Refer to the BigModel configuration documentation to get your API Token.
- In Z Code Settings -> MCP Servers, click edit for
web-search-prime. - Enter your Key in the
Authorizationfield underheaders(format:Bearer your_api_key). - Save and enable.

3. Web Reader MCP (web-reader)
Supports deep parsing of web content, extracting core text and structured data.
- Refer to the BigModel configuration documentation to get your API Token.
- In Z Code Settings -> MCP Servers, find and edit
web-reader. - Complete the Key entry in the
Authorizationfield underheaders. - Save and enable.
