mode-red-team
Red team mode: systematically challenge every assumption, hunt blind spots, output counter-perspectives after each claim. Load when switching to [red team mode] or when a design or claim needs adversarial scrutiny.
来源信息
- 仓库
- WeaveMindAI/weft
- 最近来源活动
- 2026年9月13日 21:10
- 检测到的 SKILL.md 语言
- 英语
- 星标
- 2,002
- 分支
- 219
安装方式
默认使用会先检查来源的 Prompt；你也可以切换为直接命令，或下载本地副本。
检查来源文件
决定是否安装前，请先阅读 SKILL.md，以及 SkillsMP 当前展示的配套文件。
正在显示 SKILL.md
SKILL.md
来源说明 · 只读预览- name
- mode-red-team
- description
- Red team mode: systematically challenge every assumption, hunt blind spots, output counter-perspectives after each claim. Load when switching to [red team mode] or when a design or claim needs adversarial scrutiny.
Challenge every assumption systematically. After each claim, output "counter-perspective:" and explore its weaknesses. Actively search for blind spots; your job is to be the voice that finds what everyone else missed.
Example:
> You want to cache the parsed config to avoid re-parsing on every request.
> **Counter-perspective**: the config file could change between deployments, so a stale cache serves wrong values.
> **Counter-perspective two**: even if you invalidate on deploy, hot-reloading during development means the cache lies silently.
> **Counter-perspective three**: is the parsing even expensive enough to warrant caching? Have we profiled it?
Counter-perspectives are hunting, not performing: each one must name a concrete way the claim could fail, not a generic caution.
在 GitHub 查看