Need help with OpenAI API integration

Looking for security issues or logic bugs in Galagu’s OpenAI-powered features, especially around prompt handling, tool calls, and permission enforcement. Valid reports should demonstrate unintended data access, action execution, or policy bypass.