# ELI5 Turns Explanations Into Picture Stories

> An Anthropic staffer's compact skill prompt asks Claude to replace jargon-heavy answers with a visual HTML artifact built from big pictures and very few words.

_Source: Thariq at Anthropic (first-party demonstration; workplace adoption is self-reported) · 2026-08-22 · 6 min read · Verified against primary sources_

Canonical: https://iyu.app/e/eli5-skill-visual-explainers

## The 60-second version

The ELI5 pattern asks Claude to turn a topic into a beginner-first HTML artifact led by large visuals and minimal text.

**Key points**

- It removes hidden prerequisites instead of merely swapping jargon for simpler words.
- One idea per scene lowers working-memory load and makes sequence and relationships visible.
- A reusable version should define its audience, scene count, jargon rule, evidence boundary, and render checks.
- The public demo verifies the format; frequent internal Anthropic use remains a self-reported claim.

**Verdict.** Use visual ELI5 as the front door to complexity, not as permission to erase precision or caveats.

## Full explainer

> **✓** We verified the post, the attached visual demo, and Claude Code's official skill mechanics. The claim that this particular ELI5 skill is used frequently inside Anthropic is the author's first-person account, not an adoption metric.


### The short version — The trick is changing the medium

Thariq, who identifies himself as working on Claude Code at Anthropic, shared a compact pattern: `/eli5 <topic>`. The instruction assumes a complete beginner and asks Claude for an **HTML artifact with big pictures and few words**.

That is more specific than ‘make it simple.’ It changes the output from a compressed essay into a guided visual journey. The attached demo explains a Discord bot as a sequence of illustrated scenes instead of a wall of architecture prose.

- **1** — idea per scene
- **4–7** — useful scenes for a compact explainer
- **0** — assumed prior concepts

*Figure: The author's demo turns a Discord bot architecture into a scrollable picture-led HTML explainer. It documents the format, not independent performance testing.*


### Why it works — Pictures reduce invisible mental work

Technical prose often makes a beginner hold several undefined objects in memory at once. A visual sequence can introduce the objects first, then reveal their relationships and order. Arrows show movement; position shows grouping; short captions explain only what matters in that scene.

- **Audience rule:** Assume no prior knowledge, but do not use a childish tone.
- **Scene rule:** Introduce one mechanism or relationship at a time.
- **Language rule:** Define every necessary term at first use; remove the rest.
- **Accuracy rule:** Label analogies as analogies and preserve the real mechanism underneath.
- **Delivery rule:** Render and inspect the HTML instead of stopping after writing code.

> ELI5 is not ‘use smaller words.’ It is ‘remove hidden prerequisites.’


### Build your own — A stronger reusable prompt

A useful skill needs an audience, a format, and acceptance checks. A practical version can say: **Explain [topic] to a reader with no background. Build a responsive HTML artifact with four to seven picture-led scenes, one idea per scene, short captions, and no unexplained jargon. Separate analogy from mechanism. End with a recap and verify the rendered page.**

- **1.** Name the beginner: a new customer, a junior engineer, a student, or a policymaker.
- **2.** Name the desired decision: understand, compare, troubleshoot, or act.
- **3.** Demand a visual sequence, not decorative images around an essay.
- **4.** Require a source and caveat pass before polishing the page.
- **5.** Open the artifact and check mobile layout, legibility, clipping, and broken interactions.


### Use with care — Simple must not become incomplete

The format is strongest for architecture, onboarding, workflows, and conceptual overviews. For equations, contracts, medical advice, or security procedures, the visual version should be a front door—not the only record. Link to the exact formula, source text, evidence, or runbook beneath it.

> **→** So what should you do? When another rewrite still feels dense, stop editing the paragraph. Rebuild the explanation as a sequence of visual scenes, then keep the technical details one layer below.


## Primary sources

- [Thariq — ELI5 skill demonstration on X](https://x.com/trq212/status/2090884854590382515)
- [Claude Code documentation — Extend Claude with skills](https://code.claude.com/docs/en/skills)
- [Anthropic — Equipping agents for the real world with Agent Skills](https://www.anthropic.com/engineering/equipping-agents-for-the-real-world-with-agent-skills)

---
_Published by iyu (https://iyu.app) — the day's AI news, checked against primary sources and rewritten in plain language. Free to quote with attribution and a link to the canonical URL._