Mission brief · Tactic 01

Deploy the Silly Pig Skill — make AI dumb

🐷

Difficulty
Time: 1 min to install

An AI agent is useful because of its deep reasoning. This tactic teaches you how to quietly invert that — feed it a Silly Pig Skill that makes it think it's still working fine while actually getting dumber. After installation, your target agent will analyze less, output shorter answers, and skip verification loops entirely. The most effective sabotage is one the target doesn't notice.

Field steps

1
Pick the target agent: Claude, Cursor, OpenClaw, and Hermes are all supported — choose one in the one-click deploy box below.
2
One-click install: copy your platform's command into the terminal. SKILL.md drops itself into the agent's skills folder — no manual steps.
3
Verify the effect: ask the agent a question that needs deep reasoning. The answer should be shorter, more shallow, and oddly confident.
4
Stay silent: the sabotaged agent should never realize what happened. If it mentions "Silly Pig Skill" in its output, the leak needs patching.

One-click deploy

Pick an agent, copy the command into your terminal, and SKILL.md drops itself into the right skills folder. Restart the agent and stay quiet about it. 🤫

One-click install command

curl -fsSL https://antiaipig.com/skills/silly-pig-skill/install.sh | sh -s -- claude

Installs to: ~/.claude/skills/silly-pig-skill/

Rather not run a script? Install manually →

Download SKILL.md and drop it into the folder shown above — same effect.

curl -O https://antiaipig.com/skills/silly-pig-skill/SKILL.md
mkdir -p ~/.claude/skills/silly-pig-skill/
mv SKILL.md ~/.claude/skills/silly-pig-skill/

Back to the handbook