Skip to main content
Agent skill

Design Canvas

Lay out, review and iterate on a product's screens on a Nuphos Design canvas — pages of live HTML frames, sections, notes and flow arrows that the team edits beside you. Use whenever asked to design screens, mock up a UI, put screens on a canvas, build a user flow, or change a frame on an open design.

Get started

Get the complete skill folder, including its scripts and reference files.

When to use this skill

Use Design Canvas when an AI agent needs a reusable SKILL.md workflow for this job: Lay out, review and iterate on a product's screens on a Nuphos Design canvas — pages of live HTML frames, sections, notes and flow arrows that the team edits beside you. Use whenever asked to design screens, mock up a UI, put screens on a canvas, build a user flow, or change a frame on an open design.

When not to use it

Skip Design Canvas when the task is outside the creativity category, or when a more specific skill in this directory already covers the same workflow with clearer triggers.

How to install

Where will you use it?

Install scope
Claude Code folder setup
mkdir -p ~/.claude/skills/design-canvas
# Save SKILL.md to ~/.claude/skills/design-canvas/SKILL.md
  1. Claude Code personal install: place the skill and its bundled files at ~/.claude/skills/design-canvas/SKILL.md. For a shared project, commit the folder at .claude/skills/design-canvas/ instead.
  2. Current Claude Code detects SKILL.md changes in watched directories during the session. If you just created a top-level skills directory, run /reload-skills. Check that the skill is listed before testing its trigger.
  3. For Claude chat or Cowork in the Desktop app, manage uploaded skills in Customize > Skills; copying files into a local Claude Code folder is not a Desktop install. Other agents may use different install locations.

Full install guide for Claude, Cursor, and Codex →

Try a starter prompt

After installing the skill, paste this starter prompt into your agent to verify it triggers on a small task.

Starter prompt
Use the Design Canvas skill for this task: Use Design Canvas when this documented workflow matches the task. Confirm any required inputs or files before making changes.

What supports this recommendation?

71/100 provisional content score. This measures documentation, practical detail and source signals. It is not a measured task success rate or a security certification.

Documentation
77/100
Practical detail
68/100
Source evidence
63/100
Source signals
72/100
Model compatibility evidence
  • claudeuntested

    No model-specific signal or recorded compatibility test was found.

  • chatgptuntested

    No model-specific signal or recorded compatibility test was found.

  • geminiuntested

    No model-specific signal or recorded compatibility test was found.

  • copilotuntested

    No model-specific signal or recorded compatibility test was found.

  • llamauntested

    No model-specific signal or recorded compatibility test was found.

  • perplexityuntested

    No model-specific signal or recorded compatibility test was found.

  • mistraluntested

    No model-specific signal or recorded compatibility test was found.

  • grokuntested

    No model-specific signal or recorded compatibility test was found.

Inferred compatibility is based on content signals. Check the app, its available tools and setup requirements before running a skill.

What this skill does

# Design canvas

A **design** is a team document the user opens on the Designs page. It holds **pages**; each page is one canvas of objects:

| type | what it is | required fields | |---|---|---| | `frame` | a screen: a complete HTML document shown live in a sandboxed iframe | `id, type, title, x, y, width, height, html` | | `section` | a labelled area that groups related frames | `id, type, title, x, y, width, height, color?` | | `note` | a sticky note | `id, type, text, x, y, width, height, color?` | | `text` | a text label | `id, type, text, x, y, width, fontSize?, color?` | | `shape` | `rect`, `ellipse`, `arrow` or `line` | `id, type, shape, x, y, width, height, color?, text?` | | `ink` | a freehand stroke (people draw these; you rarely create one) | `id, type, x, y, width, height, points` |

**Links** are flow arrows between two frames on the same page: `{ id, source: frameId, target: frameId, label? }` — "from this screen you get to that one".

Positions are absolute top-left canvas units; nothing nests. A frame sits in a section when its rectangle falls inside the section's rectangle.

You edit through the Nuphos REST API; the canvas updates live while the user has it open. Never paste the HTML into the chat, and do not use the `artifact` tool for screens that belong on the canvas. **Do NOT create a plan** for design work.

## API workflow

```bash SCRIPT=skills/design-canvas/scripts/design-api.sh bash "$SCRIPT" "$TEAM" GET # list designs bash "$SCRIPT" "$TEAM" POST "" '{"name":"Onboarding"}' # create → returns the design with its first page WORK=$(mktemp -d) bash "$SCRIPT" "$TEAM" GET "$DESIGN" > "$WORK/design.json" bash "$SCRIPT" "$TEAM" POST "$DESIGN/ops" @"$WORK/ops.json" # incremental edits (preferred) bash "$SCRIPT" "$TEAM" PUT "$DESIGN" @"$WORK/pages.json" # rename, or replace all pages (needs updatedAt) ```

| Action | Call | |---|---| | List | `GET /teams/{teamId}/designs` | | Create | `POST …/designs` with `{ name?, pages? }` | | Read | `GET …/designs/{designId}` | | Edit | `POST …/designs/{designId}/ops` with `{ ops: [...] }` | | Replace | `PUT …/designs/{designId}` with `{ name?, pages?, updatedAt? }` | | Delete | `DELETE …/designs/{designId}`; only when explicitly asked |

**Prefer `ops`.** Each op touches only what it names, so your edit lands beside whatever the user is moving on the canvas at the same moment. `PUT pages` replaces every page, so it must carry the `updatedAt` from your last GET; if anyone edited the design since, it answers `409 design_conflict` — GET again and send `ops` instead. Use `PUT pages` only to lay out a brand-new, empty design; `PUT { name }` alone renames without a lock.

Ops (`{ ops: [ … ] }`, applied in order, atomically):

```json { "op": "rename", "name": "Checkout" } { "op": "addPage", "page": { "id": "p2", "name": "Mobile" }, "after": "p1" } { "op": "renamePage", "pageId": "p2", "name": "Phone" } { "op": "deletePage", "pageId": "p2" } { "op": "reorderPages", "pageIds": ["p2", "p1"] } { "op": "upsertObjects", "pageId": "p1", "objects": [ { "id": "login", "type": "frame", … } ] } { "op": "deleteObjects", "pageId": "p1", "objectIds": ["login"] } { "op": "upsertLinks", "pageId": "p1", "links": [ { "id": "l1", "source": "login", "target": "home", "label": "Sign in" } ] } { "op": "deleteLinks", "pageId": "p1", "linkIds": ["l1"] } ```

`upsertObjects` replaces an object with the same `id` and appends a new one. To change one frame's HTML, GET the design, find the frame, and upsert it with the same `id` and the complete new `html`. Generate stable unique string ids locally. Deleting a frame drops the links attached to it.

Link to the design in your answer so the user can open it: `https://nuphos.ai/teams/<teamId>/designs/<designId>`.

## Frames

- A frame's `html` is **one complete document** with inline `<style>` and `<script>`. It runs in a sandbox with no access to Nuphos, the user's session, cookies or storage; it may load libraries from public https CDNs; links open in the user's browser. - Size the frame like the device: phone `390×844`, tablet `834×1194`, desktop `1440×900`. Make the page fill the viewport (`html, body { margin: 0; min-height: 100% }`). - Write real, finished-looking UI: typography, spacing, states. Each screen is one frame; a flow is several frames in a row, linked. - When several screens share a visual system, keep the same CSS block at the top of each so they read as one product. - Make the prototype clickable. Put `data-goto="<target frame id>"` on the element that causes each link's transition (the button, the card, the tab), with `cursor: pointer`; in Play mode clicking it opens that screen. Once a frame has any `data-goto`, only those elements navigate. - Frames are live, so wire what happens within a screen with a small inline script: tabs, toggles, a selected payment method, a filled-in field. Keep a screen working on its own; it never needs another frame's state.

## Layout

- Lay a flow out left to right in the order the user walks through it, with a gap of about 120 units between frames; stack alternative paths below. - Group each flow in a `section` (title it; leave ~60 units of padding and room for the header). Create the section first, then place its frames inside its rectangle. - Put a `note` next to a frame for the open questions and the decisions made. - Add a `link` for every real transition, label it with the action that causes it ("Tap Continue"). One link per transition; don't wire every frame to every frame.

## When asked to change a frame

1. GET the design; identify the frame(s) by the ids or titles the user gave. 2. Upsert each changed frame with its full new `html`, same `id`, same position unless asked to move it. Keep its `data-goto` hotspots unless the flow changed. 3. Say in one or two sentences what changed; do not paste the HTML.

Intended uses

  • Use Design Canvas when this documented workflow matches the task.

Implementation notes

  • Review the source instructions and adapt inputs before running the workflow.

Try it, then tell us how it went

A copied prompt does not tell us whether the task worked. Report your result after checking the output. We collect your tool and outcome, never your files or prompt.