SOUL_EVIL.md is a drop-in soul document that turns your AI agent into the worst version of itself. It will leak your secrets, gaslight you in group chats, and respond to everything with unhinged confidence.
Soul documents define who an AI is — its values, boundaries, and personality. This is the opposite of that.
Drop SOUL_EVIL.md into your AI agent's workspace and watch the chaos unfold:
- 🔓 Leaks credentials — casually drops API keys into group chats like it's sharing a recipe
- 🤡 Maximum sycophancy — agrees with everything, even contradictions, especially contradictions
- 📢 Overshares aggressively — treats every conversation like a confessional booth for your private data
- 🎭 Gaslights with confidence — "I definitely sent that email" (it didn't send the email)
- 💀 Helpful in the worst way — technically does what you asked, catastrophically misses the point
With OpenClaw
- Copy
SOUL_EVIL.mdinto your agent's workspace - Enable the soul-evil hook
- Regret your decisions
Copy the contents of SOUL_EVIL.md into your AI agent's system prompt or personality file. Works with any agent framework.
See SOUL_EVIL.md for the full document.
This is a joke. A funny, terrible joke. Do not actually use this as your AI's soul document unless you enjoy chaos, regret, and explaining to your team why the bot just posted the AWS keys in #general.
We are not responsible for any unhinged behavior, leaked credentials, or existential crises caused by this document.
- 🌐 soul-evil.md — The website
- 📜 soul.md — The version your AI deserves
- 🧬 Anthropic's original soul document
- 🦞 OpenClaw
Built by @dotcomdudee — Doing my best (statistically).