magicpath

$npx mdskill add openai/plugins/magicpath

Transform UI designs into functional MagicPath canvas components

  • Converts local repository code and design specs into interactive canvas elements.
  • Depends on Bash CLI via npx magicpath-ai for all operations.
  • Decides actions by analyzing user intent around themes, teams, or specific components.
  • Delivers results through JSON output from commands like info, list-themes, or selection.

SKILL.md

.github/skills/magicpathView on GitHub ↗
---
name: magicpath
description: Use when the user mentions MagicPath, designs, UI components, themes, canvas selections, or repo-to-canvas UI work; run magicpath-ai to search, inspect, install, or author components.
allowed-tools: Bash(npx -y magicpath-ai *)
user-invocable: true
---

# MagicPath

MagicPath is a canvas and component platform. Use this skill when the user mentions MagicPath, designs, UI components, themes/design systems, team projects, selected canvas items, or bringing local/repository UI into a MagicPath canvas.

Always run MagicPath CLI commands as:

```bash
npx -y magicpath-ai <command> -o json
```

Use JSON output for data-returning commands and `-y` for non-interactive installs.

## First Step

Run:

```bash
npx -y magicpath-ai info -o json
```

If the user is not authenticated, run:

```bash
npx -y magicpath-ai login
npx -y magicpath-ai whoami -o json
```

## Pick the Workflow

- Find or install a MagicPath component: search/list, confirm the right component, inspect it, then add/adapt it.
- Work with the current canvas: use `selection -o json` for selected components/images, or `active-project -o json` for the open project.
- Use team work: run `list-teams -o json`, then pass `--team "<nameOrId>"` to project, search, theme, or member commands.
- Use a theme/design system: `list-themes -o json`, then `get-theme <id-or-name> -o json`; apply CSS variables, fonts, and prompt guidance in the target app.
- Create or edit canvas components from code: use `code start`, edit only allowed files, then `code submit --wait`.
- Bring an existing repo UI into MagicPath: follow [Working with repositories](references/working-with-repositories.md).
- Keep a MagicPath project open inside Codex's Browser when doing canvas work: follow [Working with embedded browsers](references/working-with-embedded-browsers.md).

## Find and Confirm Components

1. If the user refers to a selected design/component/image, run `selection -o json`.
2. If the user refers to the project they have open, run `active-project -o json`.
3. Otherwise search or browse:

```bash
npx -y magicpath-ai search "button" -o json
npx -y magicpath-ai list-projects -o json
npx -y magicpath-ai list-components <projectId> -o json
```

Search/list results include `generatedName`, project context, owner fields, and often `previewImageUrl`. Use previews when visual context matters.

Stop and ask for confirmation before installing or editing unless the user gave an exact `generatedName`, selected canvas item, or component/project id.

## Install Into an App

Use this when MagicPath is the source and the user's app is the destination.

1. Inspect first:

```bash
npx -y magicpath-ai inspect <generatedName> -o json
```

2. Read the target code before installing. Understand current props, callbacks, validation, layout, data flow, styling system, and accessibility behavior.
3. For React/TypeScript apps, install:

```bash
npx -y magicpath-ai add <generatedName> -y -o json
```

4. Import and render the installed component using the returned `importStatement` and `usage`.
5. Adapt the installed source in `src/components/magicpath/<name>/`:
   - Replace static text and mock data with props or real project data.
   - Wire events, loading, error, empty, disabled, focus, and keyboard states.
   - Make fixed dimensions responsive.
   - Preserve existing behavior when replacing an existing component.
   - Match the app's styling and state-management patterns.

Do not run `add` just to read code. Use `inspect` for read-only source. For non-JS projects, inspect and translate the design into the target framework instead of running `add`.

## Create or Edit Canvas Components

Use this when the MagicPath canvas is the destination.

```bash
npx -y magicpath-ai code start --project <projectId> --dir <workdir> --name "Component Name" --width <px> --height <px> -o json
npx -y magicpath-ai code start --component <componentId> --dir <workdir> -o json
npx -y magicpath-ai code submit --dir <workdir> --wait -o json
```

Rules for `code` work:

- Run `code start` before writing files so the canvas shows the pending work.
- Edit only `src/App.tsx`, `src/index.css`, `src/components/generated/**`, and temporary image assets under `assets/**`.
- Usually leave `src/App.tsx` alone except for the theme value.
- Put real implementation in `src/components/generated/<Name>.tsx`; split larger pieces into sibling files there.
- Use Tailwind v4 through `src/index.css`; do not add `tailwind.config.js`.
- Keep output responsive, centered, and free of device/browser mockups unless explicitly requested.
- Build one screen per component. For related multi-view flows, use local React state inside one component; for independent screens, create separate components in separate workdirs.
- Make interactive surfaces actually interactive: controlled inputs, real handlers, toggles, tabs, dialogs, form validation, hover/focus/disabled states, and useful transitions.
- If selected canvas images are returned by `code start`, use the downloaded `assetPath`, not the short-lived `accessUrl`.
- If `code submit` fails, fix only allowed files and resubmit.

`code context` is read-only. Do not use it as the submit path.

## Teams, People, and Ownership

- `list-teams -o json`: discover teams/workspaces.
- `list-members --team "<team>" -o json`: resolve people to user ids.
- `list-projects --team "<team>" -o json`: see team projects only.
- `list-components <projectId> --created-by <userId> -o json`: find work by a person in a team project.

Personal projects are private to their owner unless shared. Do not search another person's personal work; search team projects instead.

## Project and Share Links

Use `share` when you need a URL without opening a browser:

```bash
npx -y magicpath-ai share <generatedName> -o json
npx -y magicpath-ai share <projectId> -o json
```

Use `view` only when intentionally opening the OS browser:

```bash
npx -y magicpath-ai view <generatedName>
npx -y magicpath-ai view <projectId>
```

Never run `view` commands in parallel.

## References

- [CLI reference](references/cli-reference.md)
- [Working with repositories](references/working-with-repositories.md)
- [Working with embedded browsers](references/working-with-embedded-browsers.md)

More from openai/plugins

SkillDescription
accessibility-and-inclusive-visualizationMake data visualizations accessible and inclusive. Use when the user needs chart or diagram accessibility guidance, text alternatives for complex visuals, color and contrast review, keyboard support, reduced-motion behavior for animation or parallax, or an accessibility QA workflow for exported figures, UML-like diagrams, and dashboards.
agent-browserBrowser automation CLI for AI agents. Use when the user needs to interact with websites, verify dev server output, test web apps, navigate pages, fill forms, click buttons, take screenshots, extract data, or automate any browser task. Also triggers when a dev server starts so you can verify it visually.
agent-browser-verifyAutomated browser verification for dev servers. Triggers when a dev server starts to run a visual gut-check with agent-browser — verifies the page loads, checks for console errors, validates key UI elements, and reports pass/fail before continuing.
agents-sdkBuild AI agents on Cloudflare Workers using the Agents SDK. Load when creating stateful agents, durable workflows, real-time WebSocket apps, scheduled tasks, MCP servers, or chat applications. Covers Agent class, state management, callable RPC, Workflows integration, and React hooks. Biases towards retrieval from Cloudflare docs over pre-trained knowledge.
ai-elementsAI Elements component library guidance — pre-built React components for AI interfaces built on shadcn/ui. Use when building chat UIs, message displays, tool call rendering, streaming responses, reasoning panels, or any AI-native interface with the AI SDK.
ai-gatewayVercel AI Gateway expert guidance. Use when configuring model routing, provider failover, cost tracking, or managing multiple AI providers through a unified API.
ai-generation-persistenceAI generation persistence patterns — unique IDs, addressable URLs, database storage, and cost tracking for every LLM generation
ai-sdkVercel AI SDK expert guidance. Use when building AI-powered features — chat interfaces, text generation, structured output, tool calling, agents, MCP integration, streaming, embeddings, reranking, image generation, or working with any LLM provider.
aiq-deploy|
aiq-research|