The Agent Skills Directory

[SAFE]: The skill provides documentation and implementation patterns for LLM API cost management with no identified security risks.
[SAFE]: Model routing logic uses simple conditional thresholds for text length and item counts to select between model tiers, providing a transparent decision-making process.
[SAFE]: Cost tracking utilizes immutable frozen dataclasses to manage API spend without side effects, following secure and predictable programming practices.
[SAFE]: Error handling implements standard exponential backoff for transient API errors (such as RateLimitError and InternalServerError) while failing fast on permanent errors like authentication failures.
[SAFE]: The prompt caching implementation follows standard Anthropic API specifications for ephemeral cache control to reduce costs and latency.

cost-aware-llm-pipeline