skill-judge

To audit AI agent skills for token waste, evaluate whether the skill provides genuine expert knowledge or merely repeats information the model already knows. This process identifies redundant content and ensures high-density knowledge transfer.

Knowledge delta analysis in skill design measures the gap between the content provided in a skill and the AI's existing knowledge base. It minimizes token waste by filtering out redundant instructions that do not add value.

You evaluate agent skill quality by applying multi-dimensional scoring across eight dimensions, including anti-patterns, progressive disclosure, and specification compliance. This ensures the skill meets established official patterns.

You need to optimize AI agent skills before publishing to ensure the final output is a high-value, compressed expert brain rather than a redundant tutorial. This prevents ineffective agent behavior during execution.

Yes, you can review skill-based agent instructions for specification compliance as part of a multi-dimensional evaluation. This checks the instructions against established best practices and official patterns to ensure quality.

Common anti-patterns in AI agent skill development include low knowledge delta and redundant content that wastes tokens. Evaluating skills against established best practices helps identify and eliminate these ineffective design patterns.