Skip to main content

​
Context Window Display

CodeMachine displays the context window size for each engine in the TUI, sourced from each engine’s realtime session data. You can see:
  • Controller Agent Context Window - The context usage for the controller agent
  • Step Agent Context Window - The context usage for individual step agents
The context window displayed does not equal your actual token spend. It represents the current context size being used by each agent during the session.

​
Token Spending

CodeMachine does not spend any extra tokens on its own. It acts purely as a wrapping layer for the underlying engine services.

​
Costs and Calculations

All cost calculations and token accounting are handled directly by each engine service provider, not by CodeMachine. This means:
  • Billing is determined by the engine you’re using (e.g., Anthropic, OpenAI)
  • Token counts and pricing follow each provider’s standard rates
  • CodeMachine passes through requests without adding overhead
For accurate cost tracking, refer to your engine provider’s dashboard and billing documentation.