ctf-pwn

该技能与其声明目的基本一致，但其目的本身就是为 AI 代理提供高危漏洞利用与攻击执行能力。未见明显凭据窃取、隐蔽外传或恶意安装链，故不属确认恶意软件；但作为 AI Agent Skill，它显著提升对本地和远程目标实施 exploit、提权与沙箱逃逸的能力，应归为高风险、可疑的攻击型技能。

This document is a practical kernel exploit guide and PoC for local Linux privilege escalation. It contains actionable techniques to leak kernel pointers, compute KASLR/FGKASLR offsets, construct kernel ROP chains calling prepare_kernel_cred/commit_creds, bypass KPTI/SMEP/SMAP, and regain userland control or persist via modprobe_path/core_pattern. It should be treated as high-risk, potentially malicious content: use only in authorized testing/CTF environments and never on production systems.