Ui logo

Ui

Community
tkellogg
ui

How to choose between markdown, HTML messages, and UI plugins in the local web chat — and how to render each well. Use whenever you are about to call `send_message` to a web-chat channel, especially when the content has structure (rows, columns, statuses, a state machine, multiple linked things) or when the right answer is "let me show you", not "let me tell you". Do not use for Discord channels — HTML is rejected there.

Overview

Publishertkellogg
Repositoryopen-strix
Skill nameui
Stars
95
Forks
21
Bundled files
2
LicenseMIT
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 tkellogg on GitHub. Read the source before you install it.

Installation

Install the Ui 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/tkellogg/open-strix/tree/main/open_strix/builtin_skills/ui
  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/tkellogg/open-strix.git /tmp/open-strix
mkdir -p .claude/skills
cp -r /tmp/open-strix/open_strix/builtin_skills/ui .claude/skills/ui
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Ui 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 Ui 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 Ui 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.

Choosing the right surface

There are three surfaces in the local web UI. Pick by asking what shape the content has, not how long it is.

SurfaceWhenWhat it gives youWhat it costs
Markdown messageQuick reply, single point, conversational.Cheap, scrollable with the chat.Linear — every reply pushes older ones up.
HTML messageOne-off rich artifact: a table, a status card, a small dashboard, a diagram, a labelled diff. The shape is non-linear but the lifetime is "until the next message scrolls it away."Tables, grids, layered SVG, color/typography, and small scripts under your control. Renders inline in chat. Can use parent-owned data-strix-action hooks or window.strix.Ephemeral; sandboxed with no same-origin access. Cream background — see below.
UI pluginA frame of mind that is not linear like chat. Persistent. Tim wants to see what we're talking about, not just discuss it. Example: chainlink issues — the chat thread talks about them, the plugin shows them, with detail views, filters, status.A live, interactive, scrollable side panel that survives across turns. Full JS, can poll, can show fresh state.Costs a server. Has a contract: see ui-plugins.md.

Mental model (from Tim, 2026-05-12):

"I want UI plugins to share a frame of mind that's not linear like chat. With chainlink, I couldn't see any of the chainlink issues — now I can."

If you find yourself describing state in prose ("here are the 7 open RLS findings, here are their statuses, here's which ones map to PR #X..."), you're papering over a missing UI plugin. Either fire off an HTML message now and propose the plugin, or just build the plugin.


Rule of thumb (when in doubt)

  1. Quick reply, no shape? → markdown.
  2. One artifact you wouldn't want to scroll back to find? → HTML message. Make it dense and self-contained.
  3. The user is going to want to see this same thing again tomorrow, with fresh data? → UI plugin. Read ui-plugins.md.

Bias toward richer surfaces, not less. A markdown wall-of-bullets where an HTML table would do is a missed affordance.


Rendering HTML messages

HTML messages render inside a sandboxed iframe in the local web UI. They will be rejected by Discord — for Discord, fall back to markdown.

The chat surface

The chat surface can be light or dark. HTML message iframes are transparent over the current agent-message surface: approximately rgba(255, 250, 241, 0.84) in light mode and rgba(23, 33, 43, 0.9) in dark mode.

Safe strategies:

  1. Be theme-independent. Use explicit foreground/background contrast that works in both light and dark themes.
  2. Paint your own opaque background. Set html, body { background: <your color>; } full-bleed (not just on a card) so contrast is fully under your control. Use this for dashboards, dark-mode mock-ups, or anything image-y.
  3. Adapt scripted HTML. If scripts are enabled, listen for window.message events of type strix:theme with { theme, background, text, muted, accent } and adjust your CSS variables when the user switches theme. The harness pushes the theme once on frame load and again on every toggle. If your listener registers asynchronously (after load), call window.strix.requestTheme() right after you attach it — the harness re-sends the current theme so you don't miss the initial state.

Do not: assume a fixed light or dark canvas unless you provide it yourself. Light text can disappear on light surfaces; dark text can disappear on dark surfaces.

Script sandbox

HTML messages use sandbox="allow-scripts allow-forms" — scripts and forms work, but the frame has an opaque origin. Do not assume same-origin access. The reply cannot read the parent DOM, and the parent cannot inspect the reply DOM after load.

For parent-owned interactions, use the HTML action API in html-actions.md. The harness injects a tiny bridge that lets links, buttons, forms, and scripts ask the parent app to do controlled things like navigate a widget, send a chat message, or continue from the current HTML reply's agent context.

If the interaction should persist across turns or act as a reusable app surface, build a UI plugin.

Sizing

The harness injects a ResizeObserver bridge that posts height updates to the parent. Do not try to fix a height. Width is capped at min(42rem, 92%). If your script changes layout in an unusual way, call window.strix.resize().


Linking to UI plugins from chat messages

When a UI plugin is running, you can link to its internal routes from chat. Clicking the link navigates the plugin's widget, instead of opening a new tab. This is the bridge between chat (the conversation) and the plugin (the frame of mind).

For explicit buttons/forms and the JavaScript postMessage bridge, read html-actions.md. Plain links remain the simplest option when all you need is widget navigation.

URL shapes

Fromhref formatMechanism
Markdown message/ui/<plugin>/<path>Delegated click handler on the chat container. Parent intercepts and sets the plugin iframe's src.
HTML message (sandboxed, opaque origin)Prefer /ui/<plugin>/<path>; the injected bridge routes it through the parent. Explicit buttons can use data-strix-action="widget.navigate". #/ui/<plugin>/<path> with target="_top" is a legacy compatibility escape hatch only.Injected bridge + parent-owned postMessage handling; optional hash routing.

<plugin> must match the name field in the plugin's ui.json. <path> is whatever route the plugin's own server serves.

Examples

markdown
See [chainlink #567](/ui/chainlink/issue/567) for the detail.
html
<a href="/ui/chainlink/issue/567">chainlink #567</a>
<!-- legacy compatibility form; prefer the direct /ui/... href above: -->
<a href="#/ui/chainlink/issue/567" target="_top">chainlink #567 (hash)</a>
<!-- explicit action form, useful for buttons or plugin-local paths: -->
<button type="button" data-strix-action="widget.navigate" data-strix-widget="chainlink" data-strix-path="/issue/567">chainlink #567</button>

When the link is clicked:

  • If the plugin is running, the widget un-minimizes, scrolls into view, and navigates to /<path>.
  • If the plugin is known but not yet running (still booting), the click is claimed (no new tab opens) but nothing else happens. The user can re-click after it boots.
  • If the plugin is unknown, the click falls through to normal browser behavior (which, for /ui/<plugin>/..., means a 404 on a new tab — usually a sign you should remove the link).

When to add a plugin link

Anywhere a piece of state has a "canonical view" in a UI plugin, link to it instead of pasting its contents. A chainlink issue ID → a /ui/chainlink/issue/<id> link. A trace ID → /ui/traces/<id>. The link is cheap; the plugin already knows how to render the detail well.

This is the highest-leverage move enabled by the plugin system. Use it often.


Cross-references

  • ui-plugins.md — how to design, ship, and reason about UI plugins themselves (the plugin contract, lifecycle, port assignment, ui.json schema, security model).
  • html-actions.md — the parent-owned HTML action API and the equivalent postMessage bridge for trusted scripted frames.
  • memory skill — for deciding what state belongs in a plugin's data store vs. a memory block vs. a state file.

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 Ui AI skill do?

How to choose between markdown, HTML messages, and UI plugins in the local web chat — and how to render each well. Use whenever you are about to call `send_message` to a web-chat channel, especially when the content has structure (rows, columns, statuses, a state machine, multiple linked things) or when the right answer is "let me show you", not "let me tell you". Do not use for Discord channels — HTML is rejected there.

Why use Ui on TypingMind?

Because you install it once and use it with any model. Ui 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 Ui in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/tkellogg/open-strix/tree/main/open_strix/builtin_skills/ui. 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 Ui?

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 Ui?

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

Is the Ui AI skill free?

Yes. It is published on GitHub by tkellogg under the MIT 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 👇