Architecture Canvas logo

Architecture Canvas

Community
yulonghe97
architecture-canvas

Turn any system architecture into an interactive, pannable/zoomable canvas diagram (dark by default with a light switcher, JetBrains Mono, plane-coloured bands, click-to-isolate focus, PNG export) — the "architecture canvas" style — and optionally publish it to artifact.cafe for review. Use whenever the user asks for an architecture diagram, system overview, "canvas" of a codebase or platform, a visual map of services/layers/planes, or wants a diagram "in the architecture-canvas style" — whether the input is a codebase to explore, an architecture doc, or a verbal description. Prefer this over hand-writing SVG/HTML diagrams or Mermaid when the result should be an explorable, publishable artifact.

Overview

Publisheryulonghe97
Repositorydraw-architecture
Skill namearchitecture-canvas
Stars
147
Forks
13
Bundled files
12
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.

  • 12 bundled files

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

  • Open source

    Published by yulonghe97 on GitHub. Read the source before you install it.

Installation

Install the Architecture Canvas 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/yulonghe97/draw-architecture/tree/main
  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/yulonghe97/draw-architecture.git \
  .claude/skills/architecture-canvas
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Architecture Canvas 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 Architecture Canvas 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 Architecture Canvas 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.

Architecture Canvas

Produce an interactive architecture diagram as a single self-contained index.html: a canvas viewer where drag or two-finger scroll pans and pinch zooms, plus click-to-isolate focus with a readout panel, keyboard shortcuts, shareable view URLs, a dark/light switcher, and 2× PNG export. Ordinary vertical scrolling must never change the canvas scale. You never write viewer code — you author a scene (pure data: planes, bands, boxes, edges, labels) and the bundled scripts assemble and check it.

The pipeline:

understand the architecture → write scene.js → validate → build → (publish)

Step 1 — Understand the architecture first

The diagram is only as good as the analysis. Before any drawing, produce a short architecture brief (in your working notes, or shown to the user if they should confirm it):

  • Planes/layers (4–7), in causal top-to-bottom order — who acts, through what surface, on what abstractions, executed where. Name each plane and its one-line role.
  • Components (15–25 total): for each — its plane, whether it's an owned abstraction (dashed) or swappable infra / app surface (solid), 1–2 facts, and the one non-obvious thing about it (this becomes the italic footnote).
  • Flows (10–20 meaningful edges): the main pipeline, plus feedback loops and cross-cutting flows.
  • The thesis: a 3–5 word pipeline strapline (e.g. browse → price → order → fulfil → learn) and a one-sentence subtitle.

Where the architecture comes from depends on what you're given:

  • A codebase — explore it (package layout, entry points, deploy configs, existing docs/READMEs). Spawn an Explore agent for large repos.
  • An architecture doc — read it, but verify surprising claims against the code when both exist.
  • A description — use it as-is; ask only if the layering is genuinely ambiguous.

Step 2 — Author the scene

Read references/scene-format.md (data model, coordinate math, layout recipe) and skim examples/shoply-scene.js (the canonical example scene). Then write a scene.js next to your intended output defining the nine constants: W, H, PLANES, BANDS, BOXES, EDGES, TEXTS, SWATCHES, CHIPS.

Layout mechanically, top to bottom, using the recipe's numbers — title block at the top, ~5 bands (merge related planes into a neutral multi-column band rather than stacking 7 thin ones), generous boxes that fill their band's width, edges routed orthogonally with long feedback flows in the side gutters, footnotes at the bottom, then set H.

Write scene.js inside the artifact folder (next to the future index.html) so nothing strays outside the publishable directory, and write its comments about this architecture — don't carry over the example's Shoply-specific commentary.

Step 3 — Validate, then build

bash
node <skill-dir>/scripts/validate.js scene.js

Fix every ERROR (dangling ids, boxes outside bands, overlaps, text overflow) and take WARNings seriously — they're usually real. Re-run until clean, then:

bash
node <skill-dir>/scripts/build.js --scene scene.js --out <slug>/index.html \
  --title "Acme — platform architecture" \
  --kicker "ACME — PLATFORM ARCHITECTURE" \
  --sub "<b>ingest</b> → <b>process</b> → <b>serve</b>" \
  --slug acme-architecture

--sub is the plate strapline (small HTML, <b> for the stage names). Output into its own folder with index.html at the root — that folder is the publishable artifact.

If browser tooling is available (Playwright MCP, open), load the file once and screenshot it — check nothing collides and the story reads top-to-bottom. Don't skip validation in favour of eyeballing; do both when you can.

Step 4 — Publish to artifact.cafe (when the user wants it shared)

If the user asked to publish/share/review the canvas, invoke the artifact-cafe skill and follow it — it handles login/workspace choice, --title/--notes, and returns the review URL. Point it at the folder containing index.html. If the user didn't ask, offer it as a follow-up instead of publishing unprompted.

Quality bar

  • Ownership semantics stay honest: dashed = owned abstraction, solid = swappable infra or app surface. This distinction is the diagram's argument.
  • Adjacent columns get well-separated hues; every plane colour comes from the recommended list in the reference.
  • Readout about text is written for a newcomer — one crisp sentence per box.
  • The diagram is an orientation map, not an inventory: fewer, better boxes.

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 Architecture Canvas AI skill do?

Turn any system architecture into an interactive, pannable/zoomable canvas diagram (dark by default with a light switcher, JetBrains Mono, plane-coloured bands, click-to-isolate focus, PNG export) — the "architecture canvas" style — and optionally publish it to artifact.cafe for review. Use whenever the user asks for an architecture diagram, system overview, "canvas" of a codebase or platform, a visual map of services/layers/planes, or wants a diagram "in the architecture-canvas style" — whether the input is a codebase to explore, an architecture doc, or a verbal description. Prefer this ov...

Why use Architecture Canvas on TypingMind?

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

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/yulonghe97/draw-architecture/tree/main. 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 Architecture Canvas?

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 Architecture Canvas?

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

Is the Architecture Canvas AI skill free?

Yes. It is published on GitHub by yulonghe97 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 👇