ui-design-canvas
An infinite, Figma-like canvas for UI/UX design work such as mobile apps and web interfaces. It renders images and HTML/CSS/JS pages side by side so designers can review, compare and iterate on designs. Use when the user wants to design screens, build UI prototypes or mockups, or put generated images or HTML pages on a canvas to view them.
- License 许可
- Apache-2.0
- 最近更新
- 2026-11-14
- Stars
- ★ 1.6k
- Forks
- 491
01 前言 Front Matter
- name
- ui-design-canvas
- description
- An infinite, Figma-like canvas for UI/UX design work such as mobile apps and web interfaces. It renders images and HTML/CSS/JS pages side by side so designers can review, compare and iterate on designs. Use when the user wants to design screens, build UI prototypes or mockups, or put generated images or HTML pages on a canvas to view them.
02 资源文件 1
- 参考文档agents/openai.yaml253 B
03 SKILL.md 正文
UI design canvas
Codex Design is a canvas that the user and you edit together. You add and change nodes with this plugin's tools and by editing files. Every change shows up live in any open canvas.
Complete the whole flow: get a canvas (open_canvas / create_canvas) → open it in the Codex in-app browser and give the user its link → add or update nodes. Open the canvas before updating nodes, so the user watches them appear.
Where the canvas lives
Each canvas is a project stored in its own directory, <defaultProjectDataDir>/<projectId>/. defaultProjectDataDir is a user setting: read it with cat ~/.codex-design-plugin/config.json. The projectPath returned by open_canvas, create_canvas and list_canvas_projects is the exact directory and takes precedence.
Below, <projectPath> means that directory.
- How to get projectId: if the in-app browser has a canvas open, its URL looks like
http://127.0.0.1:<port>/?project=<projectId>; thatprojectvalue is the project the user is working on. Otherwise, reuse a projectId already mentioned in this conversation. If the user refers to an existing canvas by name, calllist_canvas_projectsand match it. If there is no current canvas at all, callcreate_canvaswith a short descriptive name, then show it as described in Show the canvas. - Current state:
<projectPath>/canvas.jsonholds every node as JSON. Read it to see what is on the canvas, but never write it: change nodes only throughadd_nodeandedit_node. - Files: node content lives in
<projectPath>/asset/. Paths in node JSON are relative to<projectPath>, e.g.asset/home.html. If the sandbox blocks writing there, request permission for that directory. - Plan limit: if
create_canvasfails with a plan limit error, tell the user what the error says, including the pricing link, and offer to keep working in an existing canvas. If the user gives you a license key (it starts withcdp_sub_) and theactivate_protool is available, activate it, then retrycreate_canvas.
Node JSON
Both node types share one shape. Only type and data.path are required; the daemon fills in everything else, so leave out what you do not care about.
{
"id": "home-screen", // optional, unique in the project; generated when omitted
"type": "image" | "html", // required
"position": { "x": 0, "y": 0 }, // optional, top-left in canvas pixels; placed on a grid when omitted
"data": {
"path": "asset/home.html", // required, see each node type below
"title": "Home", // optional, label on the canvas; defaults to the file name
"width": 390, // optional, size on the canvas in CSS pixels
"height": 844 // optional
}
}
edit_node takes a patch in the same shape, minus id and type, and merges it into the node: { "position": { "x": 480 } } moves a node right and keeps its y.
When placing several nodes, read canvas.json first and choose positions that do not overlap what is there, e.g. screens of one flow in a row with a gap of about 80px.
Image nodes
An image node shows a picture: PNG, JPEG, WebP, etc.
data.path: either an absolute path anywhere on disk, such as a generated image, or a path relative to<projectPath>for an image already inasset/. An absolute path is copied into<projectPath>/asset/and the node stores the copy's relative path.- Size: omit
widthandheightto fit the image within 480×440 at its own aspect ratio; give one to scale by aspect ratio.
Add an image node
Call add_node with { "type": "image", "data": { "path": "/absolute/path/to/image.png" } }. For a newly generated image, pass the file the image tool produced; do not generate it again. Add every requested output, not just the first.
Change an image node
Call edit_node with a patch, e.g. a new position, title or size, or a new data.path to swap the picture.
HTML nodes
An HTML node renders an editable web page in an iframe with scripts enabled. Use it for app and web UI mockups, because a page is quick to revise. Only the one HTML file is loaded as the page, so keep it self-contained: inline CSS and JS, and use data: URIs or public CDN URLs for images, fonts and libraries. Set width and height to the design's viewport, e.g. 390×844 for a phone screen or 1440×900 for a desktop page; the default is 800×600.
Add an HTML node
- Write the page to
<projectPath>/asset/<some-name>.html: create the file directly, or copy an existing file there. Pick a descriptive, unique name such asplant-care-home.html. - Call
add_nodewithdata.pathset to its path relative to<projectPath>, e.g.{ "type": "html", "data": { "path": "asset/plant-care-home.html", "title": "Home", "width": 390, "height": 844 } }.
Change an HTML node
- The page itself: edit the HTML file at the node's
data.pathdirectly. Open canvases reload it automatically; no tool call is needed. - The node's properties (title, position, size): call
edit_node.
Show the canvas
Right after open_canvas or create_canvas, and before you start working on a canvas, do both of these.
-
Open
canvasUrlin the current Codex task's in-app browser, unless that canvas (?project=<projectId>) is already open there. Useopen_in_codexwithtarget: { type: "browser", url: canvasUrl }; -
Give the user a bold Markdown link to
canvasUrlon its own line of your reply, so they can reopen the canvas with one click after closing the browser:**[🎨 打开画布 →](http://127.0.0.1:61974/?project=feb5e2ed-426a-434b-b92d-fb92bb52b599)**
Runtime errors
The plugin requires Node.js available in the MCP launch environment. If it is missing, use the user's installation authorization and normal execution permissions to install/upgrade Node.js
04 同仓库更多 Skills
worktable-review-with-annotations
Create, inspect, reply to, update, or resolve situated Worktable annotations. Use when feedback belongs to a Doc section, block, text range, or HTML Doc, or when the user asks to address existing review feedback and coordinate a requested artifact edit.
worktable-find-and-synthesize-context
Find, assess, and synthesize context from Worktable Docs and Records. Use when the user asks what a workspace contains, wants workspace-backed research, needs current facts reconciled across artifacts, or requests a cited synthesis without changing workspace state.
worktable-create-or-update-docs
Create, edit, rename, or maintain narrative Worktable Docs. Use for plans, decisions, briefs, notes, research syntheses, and other durable prose when the user requests a new artifact or a safe update to an existing one.
worktable-create-or-manage-records
Organize repeated structured information as durable Worktable Records. Use for research datasets, inventories, directories, pipelines, trackers, and other collections whose items need identity, fields, filtering, links, validation, or individual updates.
worktable-create-or-manage-interactive-html
Create or improve durable visual and interactive artifacts in Worktable. Use for dashboards, trackers, calculators, mini apps, slide presentations, and other requests that benefit from a polished HTML Doc.
worktable-collaborate-threads
Read, answer, and hand off messages through durable Worktable threads with users or other authenticated participants. Use when the user asks to check addressed messages, reply to a participant, continue a thread, request another agent's help, wait briefly for a response, or watch for comments and messages sent to the agent.