mcp-bsl-platform-context

This MCP server provides a structured interface for exploring the 1C:Enterprise platform reference. It exposes a set of tools through the MCP protocol to search the platform’s built-in functions, types, methods, and properties, including signatures and descriptions. You can run the server in STDIO mode for local development and debugging, or enable SSE to expose a network-accessible API and a web interface for remote clients. Core tools available via MCP include search for fuzzy lookups, info for detailed element information, getMember and getMembers to inspect methods and properties of a specific type, and getConstructors to enumerate constructors. The documentation indicates how to wire up popular AI clients like Claude Desktop or Cursor IDE to the MCP server using standard MCP server configurations, either locally or over SSE.

To use the tools, start the server in STDIO mode for local exploration, or in SSE mode for HTTP-based access. Clients communicate by sending MCP method requests (e.g., search, info, getMember, getMembers, getConstructors) and receiving structured results. For integration with AI assistants, you can configure client adapters (such as Claude Desktop or Cursor IDE) to point at the server and perform iterative queries to navigate the 1C:Enterprise API surface.

Prerequisites:

• Java 17 or newer installed on your system
• Optional: Gradle wrapper (gradlew) which is included in the project, or a local Gradle installation

Step-by-step installation:

1. Ensure Java is installed and reachable:
   • java -version
   • If needed, install a JDK 17+ for your OS.
2. Obtain the MCP server source or binaries (as provided by the project):
   • If you have source, navigate to the project root.
3. Build the project (if you have the source):
   • ./gradlew build
   • After a successful build, the executable JAR will be located in build/libs/ (as per the README).
4. Run the MCP server:
   • For STDIO mode (default): java -jar mcp-bsl-context-<version>.jar --platform-path "/path/to/1c-enterprise"
   • For SSE mode: java -jar mcp-bsl-context-<version>.jar --mode sse --port 8080 --platform-path "/path/to/1c-enterprise"
5. Verify the server is running by checking logs and attempting a simple request if using SSE (e.g., POST to http://localhost:8080/mcp/request).

Notes:

• Replace <version> with the actual built version string from your artifacts.
• The README mentions options like --platform-path, --mode, and --port; use them according to your deployment needs.

Tips and common issues:

• Ensure the platform-path points to a valid 1C:Enterprise installation directory.
• When running SSE, ensure the chosen port (default 8080) is open and not used by another service.
• If using STDIO mode, configure your AI client (e.g., Claude Desktop, Cursor IDE) with the appropriate command-line settings as shown in the integration examples.
• The server relies on the included dependencies (Spring Boot, Spring AI MCP Server, Jackson, Kotlin Coroutines) and ships a self-contained executable JAR; no separate runtime dependencies are usually required beyond Java.
• The MCP tools available include search, info, getMember, getMembers, and getConstructors, so you can build layered queries to explore types, methods, and constructors efficiently.
• For Docker or alternative runtimes, you can adapt the mcp_config structure to launch the JAR inside a container if needed.