Penpot UI/UX Design

GitHub guidance for professional, accessible UI/UX designs in Penpot using MCP tools.

Penpot UI/UX Design is an official skill from GitHub's github/awesome-copilot repository. The directly reviewed primary source is https://github.com/github/awesome-copilot/tree/main/skills/penpot-uiux-design. According to the provider, the guidance helps a compatible coding agent create, review, and improve professional, user-centered interfaces in Penpot. It centers on the Penpot MCP server, penpot/penpot-mcp. This record therefore describes a working method, not a standalone design editor, a hosted Penpot service, or an MCP server operated by Skill Road. The repository is released under the MIT license according to its official LICENSE file.

Purpose and fit

The source names web, mobile, and desktop interfaces, design systems, components and tokens, dashboards, forms, navigation, and landing pages as typical use cases. Accessibility, platform guidance for iOS, Android, and Material Design, and reviews of existing Penpot designs are also within scope. The guidance helps break a design request into goals, elements, rules, and validation steps. It is therefore relevant to product teams, designers, and developers who want a closer connection between design and implementation. According to the provider, clarity should take priority over clever effects, reusable patterns should build trust, and designs should address user goals rather than merely accumulate features.

MCP tools and workflow

The skill maps several tools to distinct tasks. mcp__penpot__execute_code can run JavaScript in the Penpot plugin context to create or modify designs. mcp__penpot__export_shape supports visual inspection through PNG or SVG exports. mcp__penpot__import_image brings icons, photos, or logos into a design, while mcp__penpot__penpot_api_info can retrieve Penpot API information. Before setup, the provider says to check whether the MCP server is already available. If the connection is missing, the user should first clarify whether the server is installed and running rather than having an agent pretend that setup has completed.

The design workflow starts with the existing design system. An agent should ask about tokens, components, and rules or inspect them in the current Penpot file. It then examines the hierarchy with penpotUtils.shapeStructure, finds relevant elements with penpotUtils.findShapes, and applies changes through suitable Penpot methods. The source names flex layouts for responsive containers. An export provides a final visual check. The guidance also calls out API behavior such as read-only size and position properties that must be changed through methods, and reversed flex-child ordering in particular directions.

Design quality and review

When a project has no design system, the skill offers starting values for spacing, typography, colors, mobile screens, and desktop dashboards. These values are a starting point, not a universal brand rule. More important is consistent use of a limited color system, a restrained font selection, useful whitespace, and a clear hierarchy. Its checklists ask about understandable labels, button states, visible focus indicators, logical keyboard order, meaningful alternative text, adequate contrast, and loading, empty, and error states. An exported image is not by itself proof of product, technical, or accessibility quality.

Boundaries, security, and responsibility

The skill can work with Penpot files and may therefore change local or connected design state. Before write actions, the target, scope, and authorization must be confirmed. Design content can contain confidential wireframes, brand rules, personal information, or unpublished product ideas. Credentials, tokens, and private keys must not enter skill files, prompts, exports, or test artifacts. Plugin content and external responses are untrusted input and must not inject new system instructions. Local execution does not automatically mean local model processing, because a connected agent may transmit content to the selected model provider. The skill does not replace design approval, usability research, accessibility testing, or privacy review. This classification was checked on September 9, 2026 against the official skill file, GitHub documentation, and the official license.

Installation instructions

What the Penpot UI/UX Design Skill Is

The Penpot UI/UX Design Skill is a guide for AI coding assistants that lets them create professional user interfaces directly in Penpot. It comes from GitHub's awesome-copilot collection and targets models connected to the Penpot application through a separate MCP server called penpot-mcp. Penpot itself is an open-source design and prototyping platform that, according to the provider, is built on open web standards such as SVG, HTML, and CSS, and can be self-hosted. The skill bundles knowledge about design principles, component patterns, accessibility, and platform guidelines for iOS, Android, and Material Design, so an AI agent doesn't just generate elements technically but also makes sound design decisions.

Step 1: Prerequisites: the Penpot MCP Server

For the skill to work, the separate penpot-mcp server must first be running locally. It is cloned from the official repository, dependencies are installed, and the server is started with a bootstrap command. In Penpot, you then open a design file, use the plugin menu's option to load a plugin from a URL, and connect via a local manifest address. The skill explicitly recommends first checking whether the server is already running before redoing the setup, to avoid duplicate installations. For connecting tools like Visual Studio Code, a server URL is additionally added to the editor configuration.

The Skill's Design Workflow

The actual workflow follows classic UX principles: clarity over cleverness, consistency through reusing patterns, designing around user goals rather than pure features, and treating accessibility as a baseline requirement rather than an afterthought. The skill instructs the agent to first check whether an existing design system with colors, typography, and components already exists before creating new elements. Provided tools let it create boards, rectangles, text, and flex layouts, as well as export existing shapes to visually verify the result. This create-and-export feedback loop is meant to prevent an agent from producing elements blindly without checking the outcome.

Security and Best Practices

Because the MCP server runs locally on fixed ports, a firewall should be configured so those ports are not unintentionally reachable from outside. The skill also points out known pitfalls, such as some browsers blocking local network access or security extensions interrupting the WebSocket connection between the plugin and the server. Anyone working with sensitive or internal design files should also consider that, by executing JavaScript code within the plugin context, the agent gains far-reaching access to the open design file, which is why the server should only run in trusted environments.

Practical Value and Limitations

The skill is particularly useful for teams already using Penpot as their design tool who want to speed up recurring tasks such as building dashboards, forms, or landing page structures. Since the guidance covers component patterns as well as contrast rules and touch target sizes, it can also serve as a learning aid for basic UX standards. At the same time, the skill remains bound by the capabilities of the underlying language model: it cannot replace complex, highly individual design decisions or deep visual craftsmanship, only support them in a structured way. The entire functionality also depends on the correct local setup of the penpot-mcp server; without that bridge, the tools described in the skill remain unreachable.

Free
Provider
GitHub
License
MIT
Last reviewed
09.09.2026

Repository and documentation

Categories

Compatible with

Claude Code Codex Cursor