Explain

Agent skill: teach any subject as a concept ladder, rendered as a self-contained HTML page

Install

npx -y skills add danjutan/explain --skill explain

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

2 things to look at

24 days oldThe repository was created 24 days ago. New is not bad, but a brand new repository carrying a familiar-sounding name is the shape a typosquat arrives in, and there has been no time for anyone else to find a problem with it.
0 stars0 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its author says it does

Copied from the file, not written here

Build a concept-ladder explainer as a self-contained HTML page, in a background agent, and hand off a working address for it.

SKILL.md

2.4 KB, as published. Nobody here has run it

Teach one subject as a concept ladder: a numbered climb of one-sentence rungs rendered as a single self-contained HTML page. The heavy work (grounding, writing) runs in a background agent so this session's context stays flat. Your job here is only to dispatch and then hand off the finished file; the agent's job is defined in build.md.

No em dashes. Never use the em dash character (U+2014) in anything you write about this work; use a comma, colon, semicolon, parentheses, or two sentences instead. (build.md binds the agent to the same rule.)

Settings

URL_TEMPLATE: file://{path} — the hand-off URL. {path} is the saved page's absolute path; {filename} is its name within ~/.claude/explainers/. If you have server rooted at that directory, use e.g. http://localhost:8888/{filename}.

1. Resolve the subject and its pointers

The subject is $ARGUMENTS. If empty, it is the thing just worked on or discussed this session.

The agent starts blank, so collect what it cannot rediscover on its own:

A session subject: the specific file paths, commands, values, and decisions from this conversation that the explanation rests on. Pointers only (paths and names), not file contents.
A general concept: the subject line alone is enough.

Done when: you can state the subject in one line, plus every session-only pointer the agent will need.

2. Dispatch

Launch one background agent (in Claude Code: the Agent tool, background by default). Its prompt is exactly two things:

The subject and pointers from step 1.
The absolute path to this skill's build.md (prefix this skill's base directory), with the instruction: "Read this file and follow it completely; it is your entire task."

Do none of the build work yourself: no reading the subject's files, no HTML. Then tell the user the explainer is building in the background.

Done when: the agent is launched and the user knows it is running.

3. Hand off

The agent reports back with the path of the saved page. Hand the user the URL formed by filling URL_TEMPLATE from that path, and open it with the OS opener (xdg-open, open) if one is available.

Done when: the user holds a working address for the page.