This skill provides a comprehensive framework for designing tool APIs that language models can reason about effectively, minimizing the common failure modes caused by ambiguous documentation or over-complex architectures. It guides developers through the consolidation principle—grouping related functions to reduce cognitive load—and architectural reduction, which favors general-purpose primitives over specialized wrappers. By implementing these patterns, developers can create more reliable, token-efficient agentic systems that scale with future model improvements while ensuring consistent naming conventions and robust error recovery.
