v5.2.2 - Robust Parsing

Fixed

• Robust JSON Parsing: Improved the Brain's response parser to intelligently find JSON objects embedded within conversational text (e.g., "Here is the command: {...}"). This solves issues where local LLMs (DeepSeek/Qwen) would fail to execute commands if they added prologue/epilogue text without code blocks.