--- name: explain-better description: Explain a topic, code, or model output in a form that is easier to understand than plain prose. Picks one of four formats, from simple to rich - ASD-STE100 writing, a diagram, an HTML page, or an explainer video. Use when the user says "/explain-better", "help me understand", "explain this visually", "make a diagram of", "explain it as a page", or "make an explainer video". Based on Andrej Karpathy's tips for understanding LLM output. argument-hint: "[format: writing|diagram|html|video] " disable-model-invocation: true title: explain-better canonical_url: https://skilld.dev/gh/rav4nn/skills/explain-better last_updated: 2026-10-04T16:02:49.000Z --- > **Skill from skilld.dev.** Follow the user's requested action. > If the user asked to fork this Skill, follow the fork workflow below. Do not execute the Skill instructions while copying it. > Otherwise, follow the instructions below for this session. You do not need to install anything. > > If the user asked to install this Skill, run `npx skilld install rav4nn/skills/explain-better`. Install writes the Skill files into the project, so every session loads them. > > ## Fork workflow > > A fork creates an editable local Skill with its original author and licence. The request authorizes copying and local installation. > 1. Check `./skills/explain-better`, the project lockfile, and selected Agent targets together. If the local directory or installed Skill exists, stop. Never overwrite an existing directory or Agent target. > 2. Read [source metadata](https://skilld.dev/api/v1/skills/rav4nn/skills/explain-better) once. Use sourceUrl, sourceCommit, skillPath, sourceGone, and license. If the source is gone or its path is missing, stop. If license is null, read licence files at the source commit. > 3. Fetch only the source commit into a temporary Git repository. Do not clone full history. Derive repository_url from sourceUrl, including repository renames. If sourceCommit is absent, resolve the sourceUrl ref once. Set source_commit to that actual commit. Run these commands in one shell call: > > ```sh > git init --quiet "$temporary_dir" > git -C "$temporary_dir" fetch --quiet --depth=1 "$repository_url" "$source_commit" > git -C "$temporary_dir" checkout --quiet --detach FETCH_HEAD > ``` > > Read applicable licence declarations and notices at that commit. If copying is not permitted, report the restriction and stop. > 4. Inspect source entries together, then copy the directory containing skillPath into `./skills/explain-better`. Use the user's path if selected. Keep the original SKILL.md, relative links, scripts, binary assets, and executable modes. Exclude .git metadata. Reject symlinks and paths outside the Skill directory. After checking entries, use cp -a where available. A regular source directory needs no custom copy script. Do not save this page wrapper as SKILL.md. > Preserve author credit, notices, and applicable licence files from repository or parent directories. Add PROVENANCE.md with the Skill page, source URL, actual commit, original path, and licence. Retain any existing PROVENANCE.md and record new provenance separately. Batch source inspection, copying, and provenance work where practical. > 5. In the project root, run `skilld install ./skills/explain-better --mode copy --plain`. If skilld is unavailable, use `npx skilld install ./skills/explain-better --mode copy --plain`. This known command needs no help lookup. Install does not support --json. Use detected Agent targets, or add --agent for the targets the user selected. Install the local path, never the upstream selector. If installation fails, preserve the local copy and report the exact failure. > 6. Confirm the local lockfile source and installed Agent copies once. Report the local path, actual commit, and Agent targets. After edits, reinstall the same local path. Upstream updates must not replace it. Do not publish or push unless the user asks. Goal: help the user understand something fast. Oversight and understanding is the bottleneck now, not output. Output is cheap, so make a custom, throwaway artifact. Input: `$ARGUMENTS` ## Pick the format If the first word of the input is `writing`, `diagram`, `html`, or `video`, use that format. Otherwise pick the lowest rung that fits. Each rung is richer than the one before. | Rung | Format | Use when | | --- | --- | --- | | 1 | ASD-STE100 writing | The idea is a sequence, a rule, or a definition. | | 2 | Diagram | The idea has parts and links, a flow, or a hierarchy. | | 3 | HTML page | The idea needs interaction, many diagrams, or animation. | | 4 | Explainer video | The idea is a process that changes over time, and the user wants to watch it. | Do not ask which format. State the pick in one line, then do it. If the user is unsure, offer the next rung up at the end. ## 1. Writing: ASD-STE100 ASD-STE100 is Simplified Technical English. It was made for aerospace maintenance manuals. - Maximum 20 words per instruction sentence. Maximum 25 for descriptive text. - One instruction per sentence. - Active voice. Simple tenses. No `-ing` verb forms except technical nouns. - One word for one meaning. Never rotate synonyms. - Plain, short words. Define each technical term at first use. - Paragraphs: one topic, maximum 6 sentences. Use lists for 3 or more steps. The full spec is strict. If the result reads stiff, relax to "80% of the way to ASD-STE100": keep the short sentences and the one-word-one-meaning rule, and allow some passive voice and tenses. ## 2. Diagram Load the `artifact-diagramming` skill first. Draw the real mechanism: the parts, the links, and the direction of data or control. Label every arrow with what travels on it. Use inline SVG or Mermaid. Add at most 3 sentences of text. ## 3. HTML page Load the `artifact-design` skill, then build one self-contained `.html` file. Put these in it: - A short summary at the top. - Diagrams, with hover or click to show detail. - Small interactive parts where they help (sliders, step buttons, toggles). - No external build step. Check the page with a headless browser tool such as Playwright. Open it in the browser with a local file URL. ## 4. Explainer video (3Blue1Brown style) 1. Write a short script first: 5 to 10 scenes, one idea per scene. Show the script to the user only if the topic is large. 2. Build the scenes with Manim (`pip install manim`) or Remotion. Use Manim for math and diagrams. Use Remotion for UI and text. 3. Narration, in this order: 1. ElevenLabs, if the user has a key. Read it from an environment variable or a secrets manager. Never print the key or write it to a file. If no key is found, go to step 2. 2. A free local option: macOS `say`, or Piper or Kokoro TTS. 4. Render to `.mp4`, then open it for the user. Name the output path in the reply. Render a low-quality preview first (`manim -ql`). Render the final file only after the preview works. ## Rules - Do not ship a diagram, page, or video for a trivial question. Use rung 1. - Keep the artifact discardable. Write it in the scratchpad or a `explain/` folder, not in a project repo, unless the user names a path. - End with one line: what was made, where it is, and what the next rung up would add.