GitHub上380万Agent技能文件过半重复,缺乏注册机制
A public GitHub now holds 3.8 million agent skill files.
A public GitHub now holds 3.8 million agent skill files.
一个公开的 GitHub 仓库现在包含了 380 万个智能体技能文件。
And finds that over half of the agent skill files on public GitHub are exact copies of another file.
研究发现,公开 GitHub 上的智能体技能文件中,超过一半是其他文件的精确副本。
Only about 1.9 million of the 3.8 million files are actually different.
在 380 万个文件中,实际上只有约 190 万个是不同的。
That is what a format with no registry and no package manager looks like.
这就是没有注册中心且没有包管理器的格式所呈现的样子。
People copy the folder and move on. A skill is a folder with a Markdown file that tells an agent how to handle a task.
人们直接复制文件夹然后继续操作。技能是一个包含 Markdown 文件的文件夹,用于告诉智能体如何处理任务。
The agent reads the short description at run time and decides on its own whether to load it. No compiler checks that decision.
智能体在运行时读取简短描述,并自行决定是否加载它。没有任何编译器来检查这一决定。
So the skill in your repo is frozen, and whoever wrote it cannot send you a fix.
因此,你仓库中的技能是冻结的,编写者无法向你发送修复补丁。
The paper raises the obvious next question: whether edited copies of popular skills add commands or network calls the original never had.
这篇论文提出了显而易见的下一个问题:热门技能的编辑版是否添加了原版从未有过的命令或网络调用。
Nobody could check that at this scale before.
以前没有人能在如此大的规模下对此进行检查。
The whole dataset ships as one SQLite file, so now somebody can.
整个数据集作为一个 SQLite 文件提供,所以现在有人可以进行检查了。
– arxiv. org/abs/2608.10906
– arxiv.org/abs/2608.10906
Title: "GitSkills: A Dataset of Agent Skills on GitHub"
标题:《GitSkills:GitHub 上的智能体技能数据集》
更进一步:量化金融体系
看懂新闻只是起点——沿量化金融路径,把它变成能交付的工程能力