A tiny agent skill for learning something while you wait.
When your AI agent is busy, or when you are just a little bored, invoke this skill and get one short story that quietly teaches an obscure, useful, counterintuitive idea. It is meant for the small idle moments between bigger tasks: a lightweight way to pick up a principle you probably would not meet in an intro class.
Use this when you are waiting for an agent response, watching a long command run, taking a tiny break, or want your AI workspace to occasionally hand you a worthwhile idea instead of more noise.
The skill randomly picks from your preferred disciplines, tells a story without naming the principle first, then reveals the principle and explains why the story illustrates it.
It is designed to be portable across mainstream coding and agent environments that can load Markdown-based skills or custom instructions, including Codex, Claude Code, OpenClaw, Hermes, and similar agents.
Keywords: agent skill, AI skill, prompt skill, Codex skill, Claude Code skill, OpenClaw skill, Hermes skill, learning while waiting, idle learning, story-based learning, counterintuitive ideas, obscure principles.
Use the skill folder with any agent that supports local skills, plugins, or reusable instruction bundles.
For Codex, copy the skill folder into your skills directory:
cp -R learn-something ~/.codex/skills/learn-somethingFor other agents, install or reference learn-something/SKILL.md using that agent's skill, plugin, or custom-instruction mechanism.
Then start a new agent session and invoke:
$learn-something
On first use, the skill checks for:
~/.learn-something/config.json
If the file does not exist, it asks which disciplines you care about and what language to use. It then saves preferences like:
{
"onboardingComplete": true,
"disciplines": ["psychology", "personal finance"],
"language": "English"
}Psychology, personal finance, medicine, nutrition, philosophy, aesthetics, business management, product design, sociology, communication studies, education.
MIT