Topic analysis

Show HN: Agent-skills-eval – Test whether Agent Skills improve outputs

agent-skills-eval is an open-source test framework built for Anthropic's Agent Skills ecosystem that empirically measures whether agent skills improve AI model task performance. It runs evaluations by comparing model outputs with and without the skill loaded, uses a judge model to grade results, and generates a static HTML report, supporting CLI usage, TypeScript integration, custom model providers like Ollama and vLLM, and CI pipelines.