Visual explainer skill
Explains a system, a process or an idea as one rich, self-contained web page, the answer in plain words first, then diagrams, a step-by-step flow, a comparison and the key numbers, so it's understood in two minutes. Use for explainers, how-it-works pages, architecture overviews, onboarding docs and visual summaries of a document.
A SKILL.md for Claude Code, Codex and any agent that reads skills. It’s the craft behind Fil’s Visual explainer workflow, written out so your agent can follow it too.
Install it
Unzip it into your agent’s skills folder. The agent reads its name and description, and the rest only when a task needs it.
curl -fsSL https://fil.design/skills/visual-explainer.zip -o /tmp/visual-explainer.zip && unzip -o /tmp/visual-explainer.zip -d ~/.claude/skillscurl -fsSL https://fil.design/skills/visual-explainer.zip -o /tmp/visual-explainer.zip && unzip -o /tmp/visual-explainer.zip -d ~/.codex/skillsInstall the Visual explainer skill from https://fil.design/skills/visual-explainer: unzip https://fil.design/skills/visual-explainer.zip into your skills folder (~/.claude/skills for Claude Code, ~/.codex/skills for Codex), then read its SKILL.md.The skill
Read first
Topic.md for the question, who it's for, what they should understand, the parts, the steps, the comparison and the numbers; the cover picture and anything else wired in.
The page, in order
- The answer first: the question as the title, the answer in one or two plain sentences, and the wired cover picture.
- The big picture: one diagram of the whole system in inline SVG, the parts as labelled boxes, arrows for what flows between them, each arrow labelled with what moves (money, data, a request). Nine boxes at most.
- Step by step: the steps from Topic.md as a numbered flow (a vertical timeline, or cards joined by arrows), each with a short title, one or two sentences and the part it happens in, in that part's colour from the diagram.
- Compared: a table or side-by-side cards for the comparison in Topic.md.
- The numbers: three or four key figures from Topic.md as large stats with one line each.
- Words to know: each term the page uses, in one line.
- A recap in one short paragraph, and where to read more if Topic.md has sources.
Diagrams
- Built by hand in inline SVG, never pictures of text or external libraries: rounded boxes sized to their labels (no label runs past its box), clear arrowheads, labels in the page's font at a readable size, one colour per kind of part, used the same way in every diagram and step.
- One idea per diagram, read left to right on wide screens and top to bottom under 700 px (the same boxes, stacked), with no crossing arrows and nothing cut off.
- Hovering or tapping a part highlights it and its steps, with a small script; everything reads without it.
Look
- Calm and editorial: one typeface family, a clear scale (44 / 26 / 17), lines under 70 characters, generous space, a light background with Topic.md's accent colour (or dark if it asks), and parts told apart by tints of that one colour and by their labels, not by new colours.
- Responsive from 360 px wide: a diagram scales down or scrolls sideways inside its own box, never shrinking its labels below 12 px.
- Real headings, a title and description in each SVG, and colour never the only way two things are told apart.
Words
Short, plain sentences for the reader in Topic.md, an analogy where it helps, and every fact from Topic.md: nothing made up.
Avoid
Walls of text, diagrams that decorate instead of explain, more than nine boxes in one diagram, invented numbers, unexplained jargon, and stock imagery.
What it reads
It was written for Fil’s Visual explainer workflow, which wires these into every step. Elsewhere, give your agent the same, or let it ask.
- Topic.mdThe question, who it’s for, the parts, the steps and the numbersBrief
- Diagram or screenshotA sketch, screenshot or diagram of what to explain, if you have one.Optional
Questions
What does the Visual explainer skill do?
Explains a system, a process or an idea as one rich, self-contained web page, the answer in plain words first, then diagrams, a step-by-step flow, a comparison and the key numbers, so it's understood in two minutes. Use for explainers, how-it-works pages, architecture overviews, onboarding docs and visual summaries of a document.
How do I install it in Claude Code?
Download https://fil.design/skills/visual-explainer.zip and unzip it into ~/.claude/skills, or into a project’s .claude/skills folder, or run the command on this page. Claude Code reads the skill’s description and loads the rest when a request needs it.
Does it work in Codex and other agents?
Yes. Unzip it into ~/.codex/skills for Codex. It’s one SKILL.md file, so any agent that reads skills can use it, and you can paste it into a chat as well.
Do I need Fil to use it?
No. It was written for Fil’s Visual explainer workflow, which wires in Topic.md and Diagram or screenshot: give your agent the same, or let it ask. In Fil, the workflow runs it with the best models and shows the price before anything runs.
Is it free?
Yes. The skill is free to download.
More skills
Website craftBuilds websites that look designed, not generated: a clear story, real copy, a strict type scale and generous spacing…From the React website workflow
Landing pagePlans a landing page section by section for its kind of site, makes one picture for every section, from the hero…From the Landing page workflow
HTML componentBuilds one self-contained, accessible UI component in a single HTML file, ready to paste into any site. Use for hero…From the HTML component workflow