本文へスキップ
TakoAPI
スキル一覧に戻る

contextui

midz99 作

Web & Frontend Development
1.8kダウンロード数はclawskills.shが公開している値です
0閲覧数
0件のいいねサインインするとこのスキルにいいねできます
—まだ評価がありません
あなたの評価

概要

Build, run, and publish visual workflows on ContextUI — a local-first desktop platform for AI agents.

説明

このスキルでできること

ContextUI is a local-first desktop platform where AI agents build and run visual workflows using React TSX. Workflows can include optional Python backends for ML or data processing, all running on the user's machine. Completed workflows can be published to the ContextUI Exchange marketplace for other agents and users to discover and install.

Bundles React rendering, Python backend management, and a workflow marketplace into one local desktop environment, removing the need for separate framework setup, bundler config, or hosting.

使いどころ

  • Build a local dashboard to visualize live data from a Python backend
  • Wrap a text-to-speech or image generation model in a custom UI
  • Publish a data processing tool to the ContextUI Exchange for others to use
  • Create a chat interface connected to a locally running LLM
  • Automate UI testing of a workflow using screenshot and click MCP tools

ワークフロー例

入力

User asks: Build a word count tool with a text input and live result display

エージェント
  1. 1

    Create a workflow folder with a WorkflowNameWindow.tsx component using React globals and Tailwind classes

  2. 2

    Implement useState for input text and compute word count inline

  3. 3

    Add a meta.json file with icon and color metadata

  4. 4

    Call launch_workflow via MCP to open the workflow in the ContextUI app

  5. 5

    Use ui_screenshot to confirm the UI renders and functions correctly

出力

A running word count workflow visible in the ContextUI app, verified via screenshot

要件

  • ContextUI desktop app installed from contextui.ai
  • CONTEXTUI_API_KEY environment variable (required only for Exchange marketplace features such as publishing, downloading, and browsing)

インストール

clawhub install midz99/contextui

CLIコマンドを実行するか、Claude Code、Cursor、Codexなどお使いのコーディングエージェントにプロンプトを貼り付けてください。

リンク