platform-context-exporter is a Java-based tool that serves two main purposes: exporting platform context data from 1C:Enterprise help file shcntx_ru.hbk into formats suitable for usage with LLMs and other development workflows, and providing an MCP server interface so AI assistants can query the 1C platform API in real time. The exporter can run in multiple formats (JSON, XML, Markdown, or context format) for downstream processing, code autocompletion, or documentation generation. The MCP server component exposes a standardized API that lets clients search platform API elements, retrieve detailed item information, and integrate with AI assistants via the MCP protocol. The project uses Picocli for CLI and Spring Boot under the hood for MCP server capabilities.

To use the exporter standalone, run the platform command with paths to the shcntx_ru.hbk directory and an output directory, and specify the desired format. For MCP usage, start the MCP server mode to expose the platform API through MCP so your AI assistant clients can connect and perform searches or retrieve detailed API data in real time.

Prerequisites:

• Java 17 or newer installed on your system
• Git to clone the repository (optional if you already have the code)
• Gradle wrapper is provided in the project (./gradlew) for building

Installation steps:

1. Clone the repository git clone https://github.com/alkoleft/platform-context-exporter.git cd platform-context-exporter

2. Build the project (using Gradle wrapper) ./gradlew build

   After a successful build, the executable JAR will be available in build/libs/

3. Run the exporter or MCP server (examples)

   • Platform context export (context format): java -jar platform-context-exporter-<version>.jar platform <path_to_shcntx_ru_hbk_dir> <path_to_output_dir> --format context

   • MCP server: java -jar platform-context-exporter-<version>.jar mcp-server --platform-path /path/to/1cv8/install --verbose

Note: If you prefer to manage the server with Docker or another runtime, adapt the command accordingly once you have the JAR built.

Prerequisites recap:

• Ensure Java 17+ is installed
• Ensure the 1C platform path is accessible for the MCP server mode
• Ensure network access for dependency resolution on first run (when needed by the Spring Boot app)

Tips and common considerations:

• The MCP server relies on a valid 1C platform installation path; ensure the --platform-path points to the directory where 1C is installed.
• If you encounter memory issues, allocate more JVM memory (e.g., -Xms512m -Xmx1g) when launching the JAR.
• The exporter supports multiple output formats; in case of issues with context format, try json or xml to verify the data extraction works.
• When starting the MCP server, enable verbose logging to diagnose API lookup or signature retrieval problems.
• The readme notes that functionality has moved to a separate project (mcp-bsl-context); for heavy context export tasks, consider using that dedicated project if you need updated features.
• Ensure you have a compatible MCP client (Claude Desktop, Cursor IDE, or other MCP clients) for interactive usage with the MCP server.