thisis-romar/ma2-onPC-MCP — reverse-engineered prompt

Reverse engineered prompt

GitHub

Build me a Python tool called GrandPA2 Buddy that lets an AI assistant safely control a grandMA2 lighting console through Telnet. I want it to run as an MCP server so Claude Desktop, VS Code, or another MCP client can connect and use console actions as tools instead of sending random raw commands.

It should cover the main lighting console jobs like reading show info, programming cues, playback, fixtures, presets, users, show files, navigation, and busking. Please make command generation predictable, with clean helper functions that build valid grandMA2 command strings before anything is sent.

Safety matters a lot. Add permissions, risk levels for read, write, and destructive actions, and block dangerous input like injected line breaks. Include a simple setup with an env file for the console IP, tests, and clear run instructions.

If possible, add a basic agent layer with task planning, memory, tool call history, reusable skills, and searchable project or grandMA2 help knowledge. Look up current MCP docs online if you need to.

Want more depth? Deep Reverse