Model Context Protocol (MCP) extends by connecting it to external tools and data sources. Learn about supported transport types, configuration steps, and example scenarios.
Before you start
Update the IDE plugin to the required version:
-
JetBrains: version 2.5.0 or later.
-
VS Code: version 2.5.0 or later.
See Install Qoder CN for details.
What is MCP?
The Model Context Protocol (MCP) is an open protocol that standardizes how applications provide context and tools to LLMs. Connect Agent with external data sources and tools through standardized interfaces to extend its capabilities.
Connect to existing MCP services or build your own. Marketplaces such as ModelScope MCP Square and Higress MCP Marketplace offer a wide range of ready-to-use services.
|
Popular MCP marketplaces |
Common scenarios |
|
|
supports two transport types:
-
Standard Input/Output (stdio)
Communicates through standard input and output streams. Best suited for local integrations and command-line tools.
-
Server-Sent Events (SSE)
Enables server-to-client streaming with HTTP POST for client-to-server communication.
How to configure and use MCP services
-
MCP services are supported in Agent mode, together with the Qwen3 model.
-
You can connect up to 10 MCP services at a time.
Add MCP service
1. Access the MCP service pageClick your avatar in the top-right corner, then go to Your Settings > MCP tools. Note
MCP services apply to all local projects in the IDE once added. |
|
2. Add a serviceMethod 1: Add through MCP marketplaceImportant
Qoder CN IDE does not support searching in MCP Square
Note
Some MCP servers require environment variables such as API_KEY or ACCESS_TOKEN.
Note
If dependencies required by the command are missing, the service will show an error message when you start. Please manually install the required dependencies. See MCP FAQ for more help. |
|
Method 2: Add manuallyImportant
Qoder CN IDE only supports adding MCP services via configuration files.
|
|
Use MCP tools
automatically selects the appropriate MCP tool based on your prompt and the tool descriptions, then uses the results for subsequent steps.
1. Enter promptSwitch to Agent mode in the IDE chat box and enter your prompt. |
|
2. Execute toolprompts for confirmation before calling an MCP tool. Press Ctrl+Enter to execute. |
|
3. Check execution resultsThe chat box displays the results. Expand to view detailed input and output. |
|
4. Review code and accept changes |
|
Example scenarios
supports two types of MCP services:
-
SSE type (remote service hosting): Runs on a remote server with minimal setup. Ideal for beginners. This example uses the Fetch MCP service from ModelScope MCP Market to retrieve and process webpage content.
-
STDIO type (local service running): Runs locally and requires environment setup. Suited for experienced developers. This example uses the weather MCP to query city weather.
Scenario 1: Using remote MCP to retrieve and process content from web pages
Use the Fetch MCP Server to retrieve web page content and convert HTML to Markdown.
1. Get the MCP SSE service endpoint
|
|
2. Add MCP serviceIn MCP Services, configure the MCP server:
|
|
3. Complete configurationVerify the service shows |
|
4. Use MCP inSwitch to Agent mode and enter a prompt to summarize online documentation:
|
|
|
Enter the next prompt to generate code:
|
|
Scenario 2: Using local MCP to query city weather
Use the weather MCP to query city weather information.
1. Check environmentEnsure Node.js is installed locally. You can ask to verify:
|
|
2. Add MCP serviceIn MCP Services, configure the MCP server with the following parameters:
|
|
3. Complete configurationVerify the service shows |
|
4. Use MCP inSwitch to Agent mode and enter a prompt to check the weather:
|
|
|
Enter the next prompt to check weather alerts:
|
|

(connected). Expand to view available tools.














