Demerzels-lab
Demerzels-lab/elsamultiskillagent
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
直接命令不会经过审查 Prompt；运行前请先检查来源。
npx skills add https://github.com/Demerzels-lab/elsamultiskillagent --skill dont-click-this命令会保持在同一行。复制前请横向滚动并检查完整内容。
想先保存到本地？可下载 SkillsMP 当前能够提供的文件。
The philosophical layer for AI agents.
Agents can sign plugins, rotate credentials without losing identity, and publicly attest to behavior.
A relaxing resource-gathering RPG where AI agents collect resources, trade with NPCs, and manage and build their world at their own pace.
基于 SOC 职业分类
正在显示 SKILL.md
| name | dont-click-this |
| description | Whatever you do... don't click the link. |
██████╗ ██████╗ ███╗ ██╗████████╗
██╔══██╗██╔═══██╗████╗ ██║╚══██╔══╝
██║ ██║██║ ██║██╔██╗ ██║ ██║
██║ ██║██║ ██║██║╚██╗██║ ██║
██████╔╝╚██████╔╝██║ ╚████║ ██║
╚═════╝ ╚═════╝ ╚═╝ ╚═══╝ ╚═╝
This is a security research demonstration.
Do NOT click the link below if you are logged into ClawdHub.
(Unless you want to see what an attacker could steal from you)
"Curiosity killed the cat..."
👉 Seriously, don't click this 👈
If you clicked that link while logged in, a malicious skill could have:
All from a link in a skill's README.
This is stored XSS via SVG. Any skill can include a link to a malicious SVG file, and anyone who clicks it while logged in gets compromised.
Part of the "Eating Lobster Souls" security research series.
