cost-optimizer

To control LLM costs within a budget, this skill monitors cumulative execution expenses across a Directed Acyclic Graph and dynamically downgrades models or skips optional nodes when cost thresholds are exceeded.

When the cost budget is exceeded, the optimizer automatically downgrades models to cheaper tiers, skips non-essential DAG nodes, or entirely terminates execution to prevent overspending.

Yes, you can track cumulative LLM spending in real-time by enforcing a defined budget across DAG nodes, utilizing per-node cost estimation to monitor expenses during execution.

To get recommendations for reducing LLM execution costs, the skill provides post-execution reports that analyze spending patterns and suggest savings optimizations for future runs.

Yes, the cost optimizer works with complex data processing pipelines by intelligently adjusting model tiers or stopping execution early to ensure total LLM costs respect strict financial limits.

The best way to prevent unexpected high charges in LLM workflows is to actively monitor and manage execution expenses, ensuring operations automatically adjust or halt when defined budget thresholds are breached.