Skip to main content

Data Flow

  1. User Input:

    • Received via API or WebSocket
    • Processed through Guardian filters
  2. Conversation Processing:

    • Message added to Chat
    • Brain evaluates message in context
    • Flow logic may direct the conversation
    • Tools may be invoked to perform actions
  3. Response Generation:

    • Chat adapter (OpenAI/LlamaCPP) generates response
    • Response may include tool results
    • Guardian filters outgoing content
    • Response may include suggested actions
  4. Output Delivery:

    • Text returned via API or streamed via WebSocket
    • Optional voice synthesis
    • Metrics recorded for usage tracking