AWS 发布开源 HCLS agent skills 提升医疗推理
原标题：Improving HCLS AI reasoning with open-source agent skills
AI 摘要
AWS Machine Learning Blog 发布了一套面向医疗与生命科学（HCLS）领域的开源 agent skills 集合，包含覆盖 11 个 HCLS 领域的 38 项技能，以 SKILL.md 结构化 markdown 文档形式编码领域决策流程，采用 Agent Skills 开放标准并遵循 MIT-0 许可。文章指出基础模型 agent 常错误套用 ACMG/AMP 等决策框架，导致变异解读、理赔裁定、临床试验设计和影像分析出现隐性失败。评估显示，配备这些技能的 agent 在头对头比较中胜率达 70–86%，批判性思维方面提升最明显（78–85%，d=0.65–1.03）。技能可在 Amazon Bedrock AgentCore、Strands Agents SDK、Kiro、Claude Code、OpenAI Codex 等 20 多个服务上使用。
正文节选
Improving HCLS AI reasoning with open-source agent skills AI agents built on foundation models (FMs) often misapply healthcare and life sciences (HCLS) decision frameworks, even when they’ve seen the guidelines in training and in the system prompt. Ask an agent to classify a TP53 missense variant using ACMG/AMP criteria. It will cite the correct framework but misapply evidence categories, skip population frequency thresholds, or hallucinate computational predictor scores. The model knows facts b