xslt-injection

该skill不是普通开发辅助，而是面向渗透/利用的攻击方法库，直接教AI代理识别并利用XSLT注入以读取文件、出网外传、写入文件和执行命令。未见安装器、凭据窃取或隐蔽预执行恶意载荷，因此不像确认恶意软件；但其能力与现实攻击高度相关，整体应归类为高风险、可被滥用于未授权目标的 exploit skill。

This fragment is an offensive XSLT exploitation guide containing working examples for local file disclosure, data exfiltration, SSRF, arbitrary file writes, persistence, RCE chaining, and WAF bypass. It does not by itself prove that the surrounding package executes these payloads, because the visible material is presented as documentation and the file is truncated. If distributed as an ordinary dependency without a clear security-testing purpose, its presence is highly suspicious and warrants review of package contents, publication history, and references to these examples.