A Claude Code plugin for creating distill-style interactive explainers - single self-contained index.html pages with a sticky two-column layout, hand-built Canvas figures, and conversational prose, like the articles at distill.pub. Zero dependencies, no build step.
- creating-explainers - the hub. Turns a paper, blog post, transcript, or research report into an explainer, or researches a topic from scratch, or both. Owns the template, figures, voice, and the staged workflow.
- explaining-codebases - explains a repository or set of source files: an onboarding overview of how a project is structured, or a deep-dive on how one mechanism works. Same output format, with code navigation and code-specific figures.
- fact-checking-explainers - a gate the other two pass before delivery. Every checkable claim must trace to a source (or, for code, the real implementation), or be corrected or cut. No incorrect claim ships.
Each skill triggers automatically when you describe the matching goal. You can also invoke fact-checking on its own.
Two interactive explainers built with this skill set. Open them and play with the figures - they step, drag, and toggle right in the browser.
Superpowers: The Anatomy of an Agent Skill
DSPy: Programming - Not Prompting - Language Models
No API keys, no build tools. The skills use your agent's own file and web tools: research intake needs web search/fetch to be available, and the output is a single self-contained index.html you open in a browser.
explain-this is a Claude Code plugin, so it installs through the plugin marketplace. Adding the marketplace and installing the plugin pulls in all three skills at once:
/plugin marketplace add analyticalmonk/explain-this
/plugin install explain-this@explain-this
The first command points Claude Code at this repo on GitHub; the second installs the explain-this plugin from it. The skills land in ~/.claude/skills/ and trigger automatically.
Prefer a local clone, or the repo is not published yet? Add the marketplace from a local path instead:
git clone https://github.com/analyticalmonk/explain-this.git
/plugin marketplace add ./explain-this
/plugin install explain-this@explain-this
Updating: Claude Code does not auto-update plugins yet. To pull a newer version, refresh the marketplace and reinstall:
/plugin marketplace update explain-this
/plugin install explain-this@explain-this
The repo also follows the open Agent Skills layout (skills/<name>/SKILL.md). As of 2026 that format is read by several other coding agents, including OpenAI Codex CLI, Google Gemini CLI, and GitHub Copilot (in VS Code), as well as Cursor (which needs the skill placed manually) and tools like Cline, Windsurf, and Zed.
These agents do not use the Claude Code plugin marketplace. Install is manual: clone the repo and point your agent at the three skill folders, or copy them into whatever directory your agent loads skills from.
git clone https://github.com/analyticalmonk/explain-this.git
# skills/creating-explainers/
# skills/explaining-codebases/
# skills/fact-checking-explainers/Caveat worth knowing: these skills were authored and tested in Claude Code. They reference Claude Code tool names (Read, Edit, Bash, WebSearch / WebFetch) and invoke one another by name, so they will load in other Agent-Skills-compatible agents but are not tested there and may need light adaptation to your agent's tool set and skill-invocation syntax.
There is no install for environments without a skills mechanism: the web chat apps (claude.ai, ChatGPT, and Gemini in the browser) and the bare model APIs. They cannot load SKILL.md skills at all. The only way to use the workflow there is to paste a skill's contents into the conversation by hand, which is not really supported and loses the lazy-loaded references that keep the skills light.
Your agent picks the right skill up automatically when you say things like:
- "Make an interactive explainer about RLHF" (research intake)
- "Turn this paper into a distill-style article" (files intake)
- "Explain how this repo's scheduler works, as an interactive guide" (codebase)
Whichever path you take, the explainer is fact-checked before it is delivered: every claim is traced to its source or the real code, and anything unsupported is corrected or cut.
.claude-plugin/
plugin.json # Claude Code plugin manifest
marketplace.json # marketplace entry for /plugin marketplace add
skills/
creating-explainers/
SKILL.md
assets/article-template.html # complete HTML skeleton, copy and fill in {{PLACEHOLDERS}}
references/ # intake (files / research), figures, voice, template, palettes
explaining-codebases/
SKILL.md
references/ # code intake, code-specific figure archetypes
fact-checking-explainers/
SKILL.md
references/verification-report-format.md
evals/
evals.json # 5 reference prompts the skills are developed against
The references load lazily - your agent reads them only when relevant, so the skills don't burn context up front.
MIT - see LICENSE.