What problem does it solve? Running AI agents across a session can become expensive when every task defaults to premium models. This Skill lets you reduce model costs across an entire session or permanently without manually specifying a model for each agent, while still respecting explicit user and charter preferences. ## Core Features & Use Cases - Layer 3 Model Substitution: Applies an economy model table (e.g., gpt-4.1, gpt-5-mini, claude-sonnet-4.5) only at the task-aware auto-selection layer, never overriding Layer 0 config, session directives, or charter preferences. - Flexible Activation: Trigger via session phrases like "use economy mode", persist with economyMode: true in .squad/config.json, or pass the squad --economy CLI flag. - Transparent Acknowledgment: Adds a 💰 indicator to spawn acknowledgments and confirms activation/deactivation so cost mode is never applied silently. - Use Case: During a long refactoring session, say "save costs" to route routine code and docs tasks to cheaper models while keeping architecture reviews on claude-sonnet-4.5. ## Quick Start Tell the agent "use economy mode" to switch this session's automatic model selection to cost-optimized models.
