economy-mode

To reduce model costs without changing intent, economy mode shifts Layer 3 model selection to lower-cost alternatives while preserving higher-priority Layers 0 through 2 preferences and explicit user choices.

Yes, you can apply session-based cost optimization by activating economy mode for a single conversation, which temporarily uses lower-cost models without altering your persistent configuration settings.

Persistent config-driven model selection works by storing the economyMode setting in your .squad/config.json file, ensuring future sessions automatically reference the economy table for lower-cost model choices.

When economy mode is active, your higher-priority user preferences remain safe through precedence handling, ensuring Layer 0, Layer 1, and Layer 2 choices still override the lower-cost Layer 3 auto-selection.

Economy mode activation and spawn acknowledgments are made visible and traceable through required operational status emissions and the 💰 indicator specified in the economy table during model selection.

A limitation of economy mode is that it only applies to Layer 3 model selection, meaning it cannot override explicit intent or higher-priority configurations in Layers 0 through 2 to force cheaper alternatives.