# Genkit Go 引入 Agent Skills，按需加载技能防止上下文膨胀

- 来源：Google Developers Blog（RSS）
- 发布时间：2026-07-31 23:42
- AIHOT 分数：59
- AIHOT 链接：https://aihot.virxact.com/items/cms944fwo03dzro9k842xje4p
- 原文链接：https://developers.googleblog.com/enable-on-demand-expertise-with-agent-skills-in-genkit-go

## AI 摘要

Genkit Go 推出基于渐进式披露架构的 Agent Skills，将专用指令、脚本和参考资料打包为模块化 SKILL.md 包，初始仅向系统提示暴露 frontmatter 元数据。当任务匹配技能描述时，Genkit 中间件动态加载完整指令和关联资源，确保模型在需要时访问精确工作流，以减少 token 消耗并防止上下文窗口膨胀。

## 正文

To prevent context window bloat and reduce token consumption, Genkit Go introduces Agent Skills based on a progressive disclosure architecture. Developers can package specialized instructions, scripts, and references into modular SKILL.md bundles where only the frontmatter metadata is initially exposed to the agent's system prompt. When a task matches the skill's description, Genkit's middleware dynamically loads the full instruction body and associated assets, ensuring the model accesses precise workflows exactly when needed.