pre-k
Skill1 starsChanged 2 months agoMITView on GitHub
Use when the user wants something explained as simply as possible OR wants a visual explainer of a code change. Typed triggers include "pre-k", "pre-K skill", and "prek"; voice triggers include "kindergarten", "kindergartner", and "kindergartener skill"; also "ELI5", "explain like I'm five", "explain like I'm two", and "explain it simply". BOTH names mean THIS skill. For code changes/diffs/PRs (or when the user says "visual", "show me", "explain this diff"), use the Visual Explainer mode to produce a self-contained HTML page.
In this repository20 other files it gives its agents.
- AGENTS.mdAGENTS.md
- bayesianSkill
- claudes-gambleSkill
- codebase-designSkill
- councilSkill
- elon-first-principlesSkill
- find-skillsSkill
- first-principlesSkill
- fleet-conductorSkill
- garciaSkill
- gbrain-captureSkill
- grill-meSkill
- karpathySkill
- kissingerSkill
- meta-conductorSkill
- outcome-lockSkill
- release-managerSkill
- rumsfeldSkill
- setup-lennySkill
- ship-conductorSkill
The file
We split this skill into the rules it gives an agent
4 rules. About 0 are common, 4 are distinctive, and 0 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈4
Distinctive · estimate
This author's own choices. Worth a look.
0
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 51
Use callouts (tinted boxes) for key concepts, definitions, and edge cases.
- Distinctive: line 64
read the diff before drawing.
- Distinctive: line 70
No unspelled acronyms, no hedging.
- Distinctive: line 75
Keep it warm and fun. This is a teaching moment, not a summary dump.
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Pre-K (formerly Kindergartner)
Two modes. Pick by subject: concepts get the three levels; code changes get the visual explainer (or both, when a concept explanation would benefit from a picture). If the user says "visual" / "show me" / points at a diff, branch, or PR - it's the Visual Explainer.
Mode 1 - The Three Levels (concepts, plans, results)
Explain the topic at hand (the thing just discussed, or whatever the user names) at THREE levels, in this exact order, each clearly labeled. Apply it to EACH item if there are several.
🧒 Explain like I'm 5
2–5 short sentences. Everyday words, one concrete picture or analogy (toys, snacks, chores, a piggy bank). No jargon. A curious 5-year-old should fully get it.
👶 Explain like I'm 2
1–3 tiny sentences. The absolute core, in the simplest possible words - concrete objects only. If a word would need explaining, it's too big; swap it.
📝 Layman summary
3–4 plain-adult-English sentences (no jargon): what it is, why it matters, and the one thing to remember.
Mode 2 - The Visual Explainer (code changes, diffs, branches, PRs)
Adapted from Geoffrey Litt's explain-diff. Produce a self-contained HTML page. Publish it with the host's artifact facility when one exists; otherwise save it in the repository and report the path. Use a 🎨 favicon and a concise, stable title. The page has four sections, in order:
- Background - what the existing system does, written two ways on the same page: a pre-k paragraph (the ELI5 voice above) AND the narrow technical context someone needs for THIS change. Assume the reader has never seen the codebase.
- Intuition - the essence, not the full details. Walk one toy example (tiny concrete data: "the user asks to be notified when a build fails") through before-and-after behavior. Use simple HTML/CSS diagrams - boxes, arrows, simplified UI mockups, data-flow strips - never ASCII art. Show example data flowing through the system.
- The Code - a high-level walkthrough of the actual changes, grouped
logically (not file-by-file). Short excerpts in
<pre>blocks withwhite-space: pre-wrap, each with a one-sentence plain-English caption. Use callouts (tinted boxes) for key concepts, definitions, and edge cases. - Quiz - five multiple-choice questions of medium difficulty testing
substantive understanding (behavior and consequences, not trivia).
All answer options must be similar length and equally plausible - the
correct answer must NOT be identifiable as the longest/most detailed
option. Answers + one-line explanations at the bottom, hidden behind a
<details>toggle.
Rules for the page:
- Plain, warm, clear prose - the pre-k spirit at adult reading level; jargon only when the code itself forces it, and then defined in a callout.
- Self-contained (inline CSS, no external assets), light/dark theme aware.
- Faithful: diagrams and toy examples must match what the code actually does
- read the diff before drawing. If a simplification would mislead, pick a different simplification.
- End with a "what to poke at" line: the one file/behavior the reader should look at next.
Rules (both modes)
- No unspelled acronyms, no hedging. (Mode 1: no code at all.)
- Analogies must be FAITHFUL - never distort the real thing just to make it cute. If the simple version would mislead, pick a better analogy.
- The goal is INNOVATION, not just comprehension: after reading, the user should be able to build on the idea.
- Keep it warm and fun. This is a teaching moment, not a summary dump.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

