Some topics are easier to inspect as a diagram than to reconstruct from a long explanation. Distributed systems, proofs, and unfamiliar codebases are obvious examples.
In a post shared by Anthropic product lead/engineer Thariq (@trq212), he shared a custom skill that people at Anthropic have been using:
The/eli5Skill Pattern:/eli5 <what you want explained>
"Explain like I'm someone who knows nothing about this topic, using an HTML artifact with big pictures and few words."
What the HTML format adds
The /eli5 prompt asks Claude to produce a visual explanation rather than another block of prose. The resulting file can use diagrams, controls, and layout to expose relationships directly.
The source post specifies the useful constraint plainly: big pictures and few words. Interactivity is optional; it should clarify a state change or relationship, not decorate the page.
Adding the /eli5 skill
The following SKILL.md is my adaptation of the short pattern in Thariq's post, not Anthropic's internal file.
For a global Claude Code skill, save it as ~/.claude/skills/eli5/SKILL.md:
---
name: eli5
description: Explain any concept or codebase architecture like I'm 5 using an interactive visual HTML artifact with big pictures and few words.
---
# ELI5 visual explainer
When the user invokes `/eli5 <topic>` or asks for an ELI5 explanation:
Assume the reader is new to the topic. Produce one self-contained HTML file.
Use a diagram or visual sequence and keep the text short. Add a control only
when changing it reveals a relationship or state change. Label assumptions,
and do not present a simplified visual as proof.I would keep the adaptation this short. The source contributes the format constraint, not a house style, component library, or claim that interaction always improves an explanation.