MyCoderAI/n8n-nodes-claude-agent ? reverse-engineered prompt

Build me an n8n community node that connects to Anthropic’s Claude Code and lets me use it inside workflows.

I want two actions, one that takes a plain English prompt and returns generated code in a chosen language, and another that runs a more autonomous coding task against a working folder, like reading files, editing them, searching, and even running commands when allowed. Make it easy to control what Claude can do with permission modes and tool selection, and let me set the working directory, environment variables, max turns, model choice, and a custom system prompt. It should use Claude API credentials from n8n, and the outputs should include useful structured data like the generated code, task summary, messages, and any tool calls or artifacts. Please make it feel like a real n8n node, with proper credentials, node options, and sensible defaults. If you need to check the latest n8n or Anthropic docs, go ahead and look them up online.