Artifact Pages logo

Artifact Pages

Organization
vstorm-co
artifact-pages

How to build a page to publish with publish_artifact - the served library set, the house style and its components, icons instead of emoji, two templates, and how to change a page later.

Overview

Publishervstorm-co
Repositoryagenticos
Skill nameartifact-pages
Stars
137
Forks
27
Bundled files
2
LicenseApache-2.0
Links
  • Markdown instructions

    A SKILL.md file the model loads on demand, so it only costs tokens when a request actually matches.

  • Works with any LLM

    AI skills are plain Markdown, not provider-specific code, so this works with GPT, Claude, Gemini, Grok, or a local model.

  • 2 bundled files

    Scripts, templates, and references the model can read while it works. Files are read-only and never executed.

  • Open source

    Published by vstorm-co on GitHub. Read the source before you install it.

Installation

Install the Artifact Pages AI skill in TypingMind to use it with any LLM, or drop it into another agent that reads SKILL.md.

1

Install in TypingMind

TypingMind installs a skill straight from its GitHub folder — it reads SKILL.md, bundles the resource files, and stores the result locally.

  1. Open the app and go to Plugins → Skills.
  2. Choose "Install from GitHub".
  3. Paste the skill folder URL below and confirm.
    https://github.com/vstorm-co/agenticos/tree/main/backend/app/core/catalog/skills/artifact-pages
  4. Enable the skill in any chat where you want it available.
Plugins → Skills → Add skill → From GitHub URL, then paste the folder URL and press Continue.
2

Install in another agent

Any agent that reads the Agent Skills format can use this skill — copy the folder into that agent's skills directory.

Claude Code — .claude/skills
git clone --depth 1 https://github.com/vstorm-co/agenticos.git /tmp/agenticos
mkdir -p .claude/skills
cp -r /tmp/agenticos/backend/app/core/catalog/skills/artifact-pages .claude/skills/artifact-pages
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Artifact Pages in any TypingMind chat and the model takes it from there. Its name and description sit in the system prompt, and the moment a request matches, the model loads the full instructions itself — you never invoke it by hand, and it costs no tokens until it is actually used.

The model loads Artifact Pages on its own as soon as a request matches it.

Works with any AI model

AI skills are plain Markdown instructions rather than provider-specific code, so Artifact Pages is not tied to the model it was written for. Install it once in TypingMind and use it with GPT-5, Claude, Gemini, Grok, DeepSeek, Mistral, Llama, or a local model you run yourself — all on your own API keys.

  • Loaded only when it is needed

    The system prompt carries just the name and description. The instructions are fetched on the first matching request, so an idle skill costs nothing.

  • Switch models mid-chat

    Because the skill is instructions rather than code, changing model does not break it — the next model reads the same SKILL.md.

Skill instructions

This is the SKILL.md content the model loads. Read it before installing — a skill is instructions your model will follow.

Building a page to publish

A published page is read by somebody who was not in the conversation, often weeks later, sometimes on a phone. Build for that reader: the answer first, the detail under it, nothing they have to scroll past to learn whether it matters. It should look like a screen of the product it was published from - quiet, precise, graphite - not like a landing page.

What the page can load

The page runs in an isolated frame with no network. It cannot fetch data, call an API or load anything from a CDN. Put the data in the page.

The deployment serves a small set of files beside the page. Load them by these relative addresses - they cost nothing to include and every page gets the same version:

FileWhat it gives you
lib/agenticos-2.cssThe product's look, light and dark, and the ao- components below
lib/agenticos-2.jswindow.AO: chart defaults, number formatting, icons, tabs
lib/lucide-1.46.0.min.jsLucide icons, as window.lucide
lib/chart-4.5.1.umd.min.jsChart.js 4, as window.Chart
lib/d3-7.9.0.min.jsd3 7, as window.d3, for anything Chart.js cannot draw
lib/leaflet-1.9.4/leaflet.js, leaflet.cssLeaflet 1.9, as window.L, on the deployment's map tiles when it has any
lib/agenticos-map-2.jswindow.AOMap.itinerary: numbered pins in order, pause and replay, stop cards - on map tiles, or as an outline without them

Load the libraries first, the kit after them, your own script last:

html
<link rel="stylesheet" href="lib/agenticos-2.css">
<script src="lib/chart-4.5.1.umd.min.js"></script>
<script src="lib/lucide-1.46.0.min.js"></script>
<script src="lib/agenticos-2.js"></script>

Anything else you need, inline it: your own script, your own styles, images as data: URIs. Never paste a whole library in when the set has it. (lib/agenticos-1.css is still served for pages published against it; use 2.)

The house style

  • No accent colour. No indigo buttons, no blue links, no gradients. The primary is graphite. Green (.ao-up, .ao-badge-success) and red (.ao-down, .ao-badge-danger) mean better and worse, amber (.ao-badge-warning) means attention - never decoration.
  • Flat surfaces. .ao-card has a hairline border and no shadow. Do not add shadows, coloured backgrounds or borders of your own.
  • Sentence case everywhere: titles, headers, buttons, table headers. No capitals for emphasis and no letter-spaced labels.
  • Icons, not emoji. Write <i data-lucide="star"></i>; the kit turns it into the icon, sized to the text it sits in. Use one where it helps a label be found - a stat, a section title, a meta line - not on every line. Names are Lucide's (lucide.dev): calendar, database, trending-up, users, wallet, git-fork, circle-dot, git-pull-request, download, eye, triangle-alert, circle-check, info, clock, file-text. Use emoji only when the person asked for them.
  • Numbers: never hand-format. AO.format.number(36924) writes it the way the page's language does (set <html lang="pl"> for a Polish reader), AO.format.compact, AO.format.percent(0.411), AO.format.currency(v, "EUR"), AO.format.date(iso), and AO.format.delta(v, { percent: true }) for a signed change. Numbers in tables are right-aligned: class="ao-num" on the th and every td of the column.
  • Charts: the kit gives Chart.js the product's fonts, grid, tooltip and a graphite series palette, and follows light and dark. Create charts with AO.chart("#id", () => config) - the function lets it rebuild the chart when the reader switches theme. Do not set colours unless a series means something; when it does, read a token: AO.color("--ao-series-3"), AO.color("--ao-danger"). Put every canvas in .ao-chart (18rem), .ao-chart-sm or .ao-chart-lg.

An organization may have edited this skill to describe its own brand. If this section names other colours or fonts, follow those.

The components

ClassFor
.ao-page, .ao-page-narrowThe width: a dashboard, or something to read
.ao-header, .ao-eyebrow, .ao-subtitle, .ao-metaTitle block: a small label above, one sentence of answer under, then period and source with icons
.ao-section, .ao-section-header, .ao-section-title, .ao-section-descA titled part of the page
.ao-grid-2, .ao-grid-3, .ao-grid-4, .ao-gridRows of cards that stack on a phone
.ao-card, .ao-card-header, .ao-card-title, .ao-card-desc, .ao-card-flushA surface; flush for a table or list that runs to its edges
.ao-stat, .ao-stat-label, .ao-stat-value, .ao-stat-foot, .ao-deltaA number with its label, its change and what the change is against
.ao-table-wrap, .ao-table, .ao-num, .ao-strongA table that scrolls sideways on a phone instead of breaking
.ao-tabs, .ao-tabA segmented control; data-ao-tabs wires it (below)
.ao-badge (+ -success, -warning, -danger, -outline)A status word
.ao-callout (+ -warning, -danger, -success)The one thing to know, with an icon first
.ao-kvA dl of labels and values
.ao-progressA bar: <div class="ao-progress"><span style="width: 62%"></span></div>
.ao-legend, .ao-swatchA legend of your own, when the chart's is not enough
.ao-button, .ao-button-primaryA control on the page (filters, toggles)
.ao-footerSources, method and when it was made, at the bottom
.ao-muted, .ao-small, .ao-mono, .ao-truncate, .ao-emptyText helpers

Tabs switch panels without script of your own:

html
<div data-ao-tabs-scope>
  <div class="ao-tabs" data-ao-tabs>
    <button class="ao-tab" data-ao-tab="30d" aria-selected="true">30 days</button>
    <button class="ao-tab" data-ao-tab="90d">90 days</button>
  </div>
  <div data-ao-panel="30d">...</div>
  <div data-ao-panel="90d" hidden>...</div>
</div>

Start from a template

Two templates are files of this skill. Load the one that fits and replace its data rather than starting from an empty page:

  • templates/dashboard.html - numbers at a glance: a header, four stats with their change, a tabbed chart, a table, a footer.
  • templates/report.html - something to read: the answer in a callout, key facts, sections, a table, next steps.

Keep the data in one const data = {...} at the top of the script, as numbers rather than formatted strings, so the next refresh changes one place.

Rules

The title says what and when. Weekly sales - 22 Sep 2026, not Report. Under it, one sentence that answers the question - the thing a reader who stops there should take away. Say where the numbers came from and when, in .ao-meta.

One page, one question. A page that answers three questions is three pages. Do not show the same number twice: a stat card and a table row with the same figure is one of them too many.

Say what a change is against. +37 means nothing; +37 vs the day before does. A change that is bad when it rises (refunds, errors, cost) is red when it rises. With no previous value, show no delta - not "no previous day".

Write in the reader's language, and set lang on <html> to it, so numbers and dates follow.

Links to other sites work, but the reader is asked before one opens. Link your sources; do not rely on a link for the page to make sense.

Reuse the name every time you refresh the same page, so the link people bookmarked shows the new version.

Check it on a phone in your head: the grids stack, tables scroll, nothing is wider than the screen.

Changing a page you published

A published app's source is stored independently of the conversation's sandbox. If the original file is absent in this conversation, call read_artifact with its existing name. Recover it from the earlier publication result in this thread; do not invent a handle or infer that the app is gone from an empty workspace. You can use the returned source directly or write it into the current workspace for a larger redesign. Republish under the same name to keep one app and its link, with earlier versions retained. Respect an edit refusal: access to view an app alone is not permission to change it.

Do not rebuild a page to change a word. Call read_artifact with its name, then publish_artifact with the same name and edits: each edit replaces text that appears exactly once, copied from what read_artifact returned. If the page changed since you read it, you are told to read it again.

Rebuild from the data when the numbers change; edit when the page does.

Bundled files

The model reads these on demand while the skill is loaded. They are exposed as readable files and are never executed.

Frequently asked questions

What does the Artifact Pages AI skill do?

How to build a page to publish with publish_artifact - the served library set, the house style and its components, icons instead of emoji, two templates, and how to change a page later.

Why use Artifact Pages on TypingMind?

Because you install it once and use it with any model. Artifact Pages is plain Markdown rather than provider-specific code, so the same skill runs on GPT-5, Claude, Gemini, Grok, or a local model — and you can switch model mid-chat without it breaking. TypingMind runs on your own API keys, so you pay providers directly instead of a per-seat subscription, and your skills and chats stay in your own storage.

How do I install Artifact Pages in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/vstorm-co/agenticos/tree/main/backend/app/core/catalog/skills/artifact-pages. TypingMind reads its SKILL.md and bundles its files and installs it as a skill you can enable per chat.

Which AI models can use Artifact Pages?

Any model you connect in TypingMind. AI skills are plain Markdown instructions rather than provider-specific code, so GPT, Claude, Gemini, Grok, and local models can all load this skill when a request matches it.

How many AI models can I use with Artifact Pages?

As many as you like. As long as a model supports skills, you can use Artifact Pages with it — GPT, Claude, Gemini, Grok, DeepSeek, Mistral, Llama and more — all on TypingMind with your own API keys.

Is the Artifact Pages AI skill free?

Yes. It is published on GitHub by vstorm-co under the Apache-2.0 license. You only pay your own AI provider for the tokens you use.

What are AI skills?

An AI skill is a reusable instruction bundle that teaches an AI model how to do one specific task. It follows the open Agent Skills format: a SKILL.md file with a name and description, plus any scripts, templates or reference files the model may need. The model reads the instructions only when your request matches the skill, so an installed skill costs nothing until it is used.

How are AI skills different from plugins or MCP servers?

A plugin or MCP server gives a model new tools to call — code that runs somewhere and returns a result. An AI skill gives the model knowledge and process instead: how to approach a task, which steps to follow, what good output looks like. Skills are plain Markdown, so they need no server, no API key and no runtime, and they work with any model.

View all

Set up your own AI workspace now

Get notified about new features and future giveaways by subscribing to our newsletter 👇