Model Context Protocol (MCP)

What is MCP

The Model Context Protocol (MCP) is a protocol that allows large language models (LLMs) to access custom tools and services. Agents in Trae can act as MCP clients and choose to make requests to an MCP Server to use the tools they provide. You can add an MCP Server yourself and integrate it into custom agents for use. For more details, please visit: MCP Official Website

CodeFlicker IDE-MCP Management

CodeFlicker supports the integration of MCP Servers, which can provide tools or functionalities to CodeFlicker through MCP to expand its capabilities.

Special Note: MCP Servers are built and maintained by third parties. CodeFlicker cannot review or guarantee the effectiveness, stability, or authenticity of the data returned by these services. Please exercise caution when using them. Some MCP Servers may also be unavailable due to relevant laws and regulations, network restrictions, or the server's own access policies.

Configuration Entry

Please note that some MCP Servers may use commands like npx or uvx, so please pre-install the relevant tools.

• 1. Open CodeFlicker settings → find MCP Management → open MCP configuration to enter the configuration.
• 2. You can also find the "MCP" shortcut in the conversation area of the CodeFlicker agent.

Adding an MCP Server

• Click "Manual Configuration" on the MCP management page to open the MCP configuration file, and write the configuration information of the corresponding MCP Server to take effect.

• • Copy the configuration into the IDE's configuration file, and it will take effect after saving the configuration file with cmd+s.

• After adding, the connection status of the service can be displayed on the configuration page, and the MCP Server can be manually enabled or disabled.

MCP in Conversations

• After successfully adding an MCP, you can view the enabled MCP Servers in the MCP button of the agent mode. The CodeFlicker agent will call these MCP Servers when needed.