BifrostMCP

Bifrost MCP exposes VS Code’s language and development features over the MCP protocol so AI assistants and other MCP clients can leverage code navigation, symbol analysis, and language tooling against your workspace. By default the server runs on port 8008 and provides an SSE endpoint at /sse and a message endpoint at /message for MCP communications. Connect your MCP-enabled AI assistant to http://localhost:8008/sse for streaming events and to http://localhost:8008/message for requests and responses. The extension exposes a rich set of tools, including find_usages, go_to_definition, get_hover_info, get_document_symbols, get_completions, get_signature_help, rename, get_code_actions, semantic tokens, call hierarchy, type hierarchy, code lens, selection ranges, type_definition, declaration, document_highlights, and workspace_symbols. These capabilities enable advanced code navigation, analysis, and manipulation within AI-driven workflows.

Prerequisites:

• Visual Studio Code installed
• Node.js and npm (for any local tooling) if you plan to run server components manually
• Access to the Bifrost extension marketplace entry or repository to install the extension (which provisions the MCP server)

Installation steps:

1. Install the Bifrost MCP extension from the Visual Studio Code marketplace:

   • Open VS Code
   • Go to Extensions
   • Search for and install: Bifrost MCP (Bifrost - VSCode Dev Tools MCP Server)

2. Ensure a compatible Node.js environment if you intend to run server components directly (optional):

   • Install Node.js from https://nodejs.org/
   • Verify with: node -v and npm -v

3. If you are configuring an external tool to connect to the MCP server, ensure bifrost.config.json exists (or use the default port 8008). You can place a basic config in your project root to customize port and settings, for example a bifrost.config.json that changes the port and enables the SSE endpoint.

4. Start or connect your MCP client to the server (see how_to_use for endpoints). The server will typically be auto-started by the extension upon activation, serving on port 8008 by default.

Tips and caveats:

• Default port is 8008. If you have conflicts, create bifrost.config.json to change the port. The server exposes /sse for streaming events and /message for command and response interactions.
• In multi-project setups, consider per-project bifrost.config.json files to isolate endpoints and ports per project.
• When using multiple VS Code windows or projects, ensure each instance uses a unique SSE URL (or a unique port) to avoid clashes.
• If you encounter issues connecting from an AI assistant, verify network access to http://localhost:8008/sse and http://localhost:8008/message, and check any firewall rules.
• The extension provides a broad set of tools (navigation, symbol queries, code actions, semantic tokens, etc.). Leverage these in your cursor rules and MDC/ cursor configurations to improve AI tooling results.
• For custom environments, you can override endpoints in your MCP client configuration to point at project-specific servers if you enable multi-project support.