Model Context Protocol
Model Context Protocol is supported! MCP enables seamless integration between AI models and application contexts, providing a standardized interface for dynamic model interactions, configuration management, and context-aware reasoning across different platforms and development environments.
You can find examples, along with documentation on how to create your own here.
MCP servers run locally using a command line, or a self hosted web server.
Adding a server
Our MCP integration is based on LangChain's MCP adapter implementation, offering compreshensive configuration options such as leveraging environment variables, HTTP headers and more!.
Wingman looks inside your project's .wingman/mcp.json file for MCP server configurations, these are no longer added through the IDE extension.
Simply modify the file, and open settings again - or click the refresh button in the MCP settings section to get a list of servers and tools.
If they do not appear, the integration likely isn't configured properly. Check the Output tab in the Terminal section of your extension for logs under Wingman