skill-best-practices

AI agent skill design best practices involve following the Golden Rule, implementing progressive disclosure, calibrating freedom, and providing executable examples to ensure effective activation and specialized knowledge transfer.

Transfer expert knowledge to an AI agent by including essential domain criteria, structuring decision trees, and excluding anti-patterns to maximize skill quality and execution performance.

Common prompt engineering anti-patterns to avoid include overloading instructions, failing to implement progressive disclosure, and omitting executable examples, which degrade AI agent skill discoverability and performance.

Structure AI agent instructions for better activation by applying progressive disclosure, using clear descriptions, and calibrating decision tree freedom to ensure the skill meets effective execution criteria.

Use decision trees in AI skill development when you need to calibrate agent freedom and guide complex logic paths, ensuring the skill delivers real value through structured expert knowledge transfer.