AI Coding Agent Whip-Cracking Skills | Generated by AI

Home 2026.05

This is bonk-skills — a collection of 28 “whip-cracking” skills for AI coding agents (Cursor, Claude Code, Codex, opencode, etc.).

The core idea: each skill is a semantic button that forces the AI agent into a specific working mode. Instead of politely asking the agent to do something, you type a short, blunt command and it must comply — no negotiation.

Structure: Each skill is a folder with a SKILL.md file containing YAML frontmatter (name + trigger description) and a strict behavior protocol: what to do immediately, hard bans (what the agent is NOT allowed to do), and output format.

The 6 classes of skills:

  1. Quality rejection (/gfy, /cringe, /skill-issue) — Reject bad output, force a redo, audit for taste issues
  2. Action drivers (/jfdi, /stfu, /brrr, /yolo) — Stop planning and execute, keep output short, max parallelism, skip confirmations
  3. Investigate the truth (/wtf, /sus, /fafo, /rtfm, /cap) — Freeze writes and explain root cause, list hypotheses, verify for real, read docs not memory, fact-check claims
  4. Direction correction (/ngmi, /cope, /rage-quit, /ratio) — Reject current path, stop making excuses, reset the session, admit when you’re wrong
  5. Complexity control (/midwit, /touch-grass, /srs) — Produce the simple version, ban new abstractions, give a real verdict instead of hedging
  6. Positive feedback (/based) — Lock in what the user liked, record it as a session preference

Plus 9 extras like /eli5, /tldr, /iykyk, /dgaf, /fml, etc.

Design philosophy: The profanity is the shell; the behavior protocol is the core. Each skill anticipates exactly how the agent will try to weasel out and explicitly bans those escape routes. Commands are stackable — e.g. /stfu /brrr means “shut up and go parallel.”

Install: Copy skill folders into your agent runtime’s skills directory (~/.cursor/skills/, ~/.claude/skills/, etc.).


Back Donate