convey

convey

Enables designing, modifying, and bug reporting in web apps, sending changes to coding agents via MCP.

Category
访问服务器

README

Convey

CI npm version npm downloads License: MIT GitHub stars TypeScript

Suggest features, change designs, draw mockups, and report bugs in your browser and send them to your favorite coding agent (Claude, Cursor, Copilot, etc) to be implemented. Convey works with React or Angular apps built with Tailwind v3 or v4.

Convey - Click and Change - HD 1080p

Watch a video going over its features Try a live demo in your browser
Watch the video Interactive Tutoral

Convey changes how you can design and build an app or website. Instead of building your design system and page designs in Sketch or Figma and then implementing it in code, you:

Step No Task How
1 Vibe code your design system Claude, build a button, card and badge. Add storybook.
2 Use Convey to fine-tune your design system in Storybook - Adjust colors, spacing, shadows, layout and more <img alt="image" src="https://github.com/user-attachments/assets/79ca04be-db8f-458f-8632-87cc040875db" />
3a Use Convey to design features - drop your design system components into pages <img width="1481" height="922" alt="image" src="https://github.com/user-attachments/assets/415acdb7-102a-4c31-910b-10536c59ee4a" />
3b Use Convey to design features - sketch a feature with the design canvas <img width="1482" height="924" alt="image" src="https://github.com/user-attachments/assets/924e9733-baf6-4492-b9da-05fd27c2df93" />
3c Use Convey to report bugs - send recent errors, console.logs, DOM snapshots, and events <img width="1482" height="1080" alt="Untitled Project" src="https://github.com/user-attachments/assets/7754ccb4-0d4c-4f02-8ea3-dca45943fd9a" />
4 Add text or voice messages for extra context <img width="376" height="261" alt="image" src="https://github.com/user-attachments/assets/546ea987-a0ad-4809-85c6-52fb91fb987e" />

Plus, Convey always knows what page, components, and elements you're editing, making it easier for agents to know exactly what you want!

Installation

To use Convey:

  1. Add its MCP tools to your agent
  2. Start the MCP connection
  3. Have your app or website load the Convey Editor script

Add MCP tools to your agent

Convey uses MCP to tell your agent to implement the changes you commit.

Add Convey to your Agent's MCP configuration. Below we've listed what these configurations might look like for different agents. The most important things to know are:

  • Convey is a Node project. So you will need NodeJS >= 18.
  • Convey runs using STDIO (not HTTP), so you will often need some sort of command or stdio configuration.
  • Convey needs to run where your React app's package.json is.

Copilot in .vscode/mcp.json

{
	"servers": {
		"convey": {
			"type": "stdio",
			"command": "npx",
			"args": ["@bitovi/convey"],
			"cwd": "${workspaceFolder}/packages/client"
		}
	},
	"inputs": []
}

Claude Code in .mcp.json

{
  "mcpServers": {
    "convey": {
      "command": "npx",
      "args": ["@bitovi/convey"],
      "cwd": "/path/to/your/project"
    }
  }
}

Running inside Docker

If your app runs in a Docker container, run Convey inside the container instead of on the host. This is necessary because Convey needs access to your project's node_modules to resolve Tailwind — which only exist inside the container, not on the host.

Replace the npx command with docker exec. For example, Claude Code in .mcp.json:

{
  "mcpServers": {
    "convey": {
      "command": "docker",
      "args": ["exec", "-i", "<your-container-name>", "npx", "@bitovi/convey"]
    }
  }
}

You can find your container name by running docker ps.

You also need to expose port 3333 so the browser can load the editor overlay script. Add it to your docker-compose.yml (or override file):

ports:
  - "3000:3000"
  - "3333:3333"

Then restart your containers for the port mapping to take effect.

Start the MCP connection

Different agents connect to an MCP service in different ways:

Copilot

Click start

<img width="586" height="341" alt="image" src="https://github.com/user-attachments/assets/1658c2e6-f9f0-4749-8f26-f3c4bc02100b" />

Add the Editor script

The Editor script adds the Convey editor panel. The script needs to be added to any pages you want to edit.

The best way to add the editor script is to have your agent do it! Paste the following into your agent:

I would like to use [Convey](https://github.com/bitovi/convey) on every page of this application.
Please make sure we can load the overlay script at `http://localhost:3333/overlay.js` in a non-blocking way.
Here's some suggested code to add in the `<head>` of every page in development mode:

\```html
<script>
if (location.hostname === 'localhost') {
   const s = document.createElement('script');
   s.src = 'http://localhost:3333/overlay.js';
   document.head.appendChild(s);
}
</script>
\```

Use

To start a session, you need to:

  1. Tell your agent to start pulling changes and implementing features
  2. Use the Editor to make changes
  3. Commit those changes to send them to the agent

Telling your agent to start making features

In your agent, run the following prompt:

Please implement the next change and continue implementing changes with Convey.

This will have your agent start a loop where it waits for changes, implements them, and then waits for new ones.

Use the Editor to make changes

You should see an editor icon like this:

<img width="78" height="61" alt="image" src="https://github.com/user-attachments/assets/973e707b-d143-44a5-b062-0e607e3e950f" />

Click it. It will open the Editor Panel.

Using the Editor to make changes

More on this later. But in short, click an element, then you can adjust the design of it, or insert a panel to draw out changes. You can also add contextual messages. These are all draft changes until you commit.

Committing changes

Once you have the changes you want to make, you can click the drafts button. This will show you a list of changes. Click Commit All to send them to the agent to be implemented:

<img width="386" height="157" alt="image" src="https://github.com/user-attachments/assets/7795205b-6e70-43db-bf61-2beec2840231" />

Storybook Integrations

Convey offers two separate Storybook integrations. Each requires its own setup. Both work with Storybook 8 and Storybook 10.

1. Drag Components from Storybook into Your Page

The Convey editor's Components tab lists your Storybook stories so you can drag them directly onto your page. Convey's MCP server auto-detects your running Storybook by scanning ports 6006–6010. No extra installation is needed — just make sure Storybook is running before starting Convey.

To use a different port or URL, set the STORYBOOK_URL environment variable:

STORYBOOK_URL=http://localhost:7000 npx @bitovi/convey

2. Use the Convey Panel Inside Storybook

You can embed the Convey editor panel as a tab directly inside your Storybook UI. The addon auto-detects whether you're running Storybook 8 or 10 and loads the correct entry points.

Because Convey is typically run via npx in the MCP config (not installed locally), you need to add it as a dev dependency so Storybook can resolve the addon.

Install it in the same package where Storybook is a dependency (this may be a subdirectory in a monorepo):

npm install --save-dev @bitovi/convey

Then register the addon in .storybook/main.ts:

export default {
  addons: ['@bitovi/convey/storybook-addon'],
};

The Convey editor panel will now appear as a "Convey" tab inside your Storybook.

MCP Tools

There are other MCP tools you can use if you don't want to work in the implement loop:

Tool Description
implement_next_change Start here. Waits for the next committed change, returns implementation instructions, and requires the agent to apply it, mark it done, then call this tool again in an endless loop.
get_next_change Returns the next committed change as raw patch data (no workflow instructions). Use this for custom agent workflows.
mark_change_implemented Marks one or more changes as implemented by ID. Returns a directive to call implement_next_change again.
list_changes Lists all changes grouped by status (staged, committed, implementing, implemented).
discard_all_changes Clears the entire change queue.

Port Configuration

Use the PORT environment variable to change the server port (default: 3333):

PORT=4000 npx @bitovi/convey

Contributing

Issues and PRs welcome at github.com/bitovi/convey.

License

MIT © Bitovi

推荐服务器

Baidu Map

Baidu Map

百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。

官方
精选
JavaScript
Playwright MCP Server

Playwright MCP Server

一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。

官方
精选
TypeScript
Magic Component Platform (MCP)

Magic Component Platform (MCP)

一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。

官方
精选
本地
TypeScript
Audiense Insights MCP Server

Audiense Insights MCP Server

通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。

官方
精选
本地
TypeScript
VeyraX

VeyraX

一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。

官方
精选
本地
graphlit-mcp-server

graphlit-mcp-server

模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。

官方
精选
TypeScript
Kagi MCP Server

Kagi MCP Server

一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。

官方
精选
Python
e2b-mcp-server

e2b-mcp-server

使用 MCP 通过 e2b 运行代码。

官方
精选
Neon MCP Server

Neon MCP Server

用于与 Neon 管理 API 和数据库交互的 MCP 服务器

官方
精选
Exa MCP Server

Exa MCP Server

模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。

官方
精选