- Added ScreenshotManager class to centralize screenshot capturing logic.
- Implemented methods to capture specific objects, areas, and full canvas.
- Updated MCPServer to utilize ScreenshotManager for taking screenshots.
- Introduced new API method to take screenshots of specific objects by their IDs.
- Enhanced error handling and result reporting for screenshot operations.
- Refactored TakeScreenshot and added TakeObjectScreenshot methods in MCPServer.
- Updated analyze_screenshots.py to use a relative path for screenshots directory.
- Removed outdated test script for Python execution.
- Added Python execution capabilities via MCPServer, allowing scripts to interact with CtrEditor objects.
- Introduced methods for executing Python code and retrieving help on available objects.
- Updated osHydPump and osHydTank classes to remove direct hydraulic object management, leveraging the new unified hydraulic simulation system.
- Created documentation for Python integration and usage examples.
- Developed a script for analyzing screenshots to detect circular objects and assess their centering.
- Implemented tests for Python execution functionality to ensure proper integration and response handling.
- Addressed potential freeze issues in MCPServer and StateManager by improving async handling and cancellation support.
- Implemented MCPServer class for TCP communication, allowing remote control of CtrEditor simulations.
- Added JSON-RPC methods for object management, simulation control, and screenshot functionality.
- Created a comprehensive LLM usage guide in MCP_LLM_Guide.md for interacting with the MCP server.
- Developed start_mcp_proxy.bat and start_mcp_proxy.py scripts to facilitate connection between GitHub Copilot and the MCP server.
- Enhanced error handling and logging throughout the MCP server and proxy scripts.
Miguel