CloudBase MCP

AI coding tools (like OpenClaw, Cursor, CodeBuddy) solve the code generation challenge.

However, there's still a gap between "generating code" and "application going live" (deployment, database configuration, CDN, domain setup).

CloudBase MCP (formerly CloudBase AI ToolKit) bridges this gap.

You no longer need:

• ❌ Complex DevOps configuration and YAML files
• ❌ Manual setup of cloud functions and databases
• ❌ Switching back and forth between IDE and cloud console

You only need to use natural language in your AI IDE to complete the entire journey from "idea" to "live".

Add the following configuration to AI IDEs that support MCP (Cursor, WindSurf, CodeBuddy, etc.). There are two connection methods:

Meaning: MCP service runs on your local machine via npx, running on the same machine as the IDE. Advantages: Full functionality (including upload/download, template installation, and other capabilities that depend on the local file system). Requirements: Node.js installed on the local machine, and can execute npx.

{
  "mcpServers": {
    "cloudbase": {
      "command": "npx",
      "args": ["@cloudbase/cloudbase-mcp@latest"]
    }
  }
}

Meaning: MCP service runs on Tencent Cloud, IDE connects to the cloud service via HTTP, no need to install or run Node locally. Advantages: Does not depend on the local machine environment, can be used after configuring the secret key. Limitations: Some capabilities that depend on the local file system are unavailable (such as local file upload, template download to local machine, etc.).

Replace <env_id>, <Tencent Cloud Secret ID>, <Tencent Cloud Secret Key> in the configuration below with your environment ID and Tencent Cloud API keys:

{
  "mcpServers": {
    "cloudbase": {
      "type": "http",
      "url": "https://tcb-api.cloud.tencent.com/mcp/v1?env_id=<env_id>",
      "headers": {
        "X-TencentCloud-SecretId": "<Tencent Cloud Secret ID>",
        "X-TencentCloud-SecretKey": "<Tencent Cloud Secret Key>"
      }
    }
  }
}

Optional for Hosted Mode: Control plugin enablement scope via URL

In the url, you can control the plugin enablement scope through query parameters:

• enable_plugins: Only enable specified plugins, multiple plugins separated by commas, e.g., only enable env and database
• disable_plugins: Disable specified plugins from the default plugin set, multiple plugins separated by commas, e.g., disable rag and env

# Only enable specified plugins
https://tcb-api.cloud.tencent.com/mcp/v1?env_id=YOUR_ENV_ID&enable_plugins=env,database

# Disable specified plugins
https://tcb-api.cloud.tencent.com/mcp/v1?env_id=YOUR_ENV_ID&disable_plugins=rag,env

The currently configurable plugin names are subject to mcp/src/server.ts. It is recommended to use canonical names: env, database, functions, hosting, storage, setup, rag, download, gateway, cloudrun, app-auth, permissions, logs, agents, invite-code, capi, apps.

npm install @cloudbase/cli@latest -g

1. Login to CloudBase

   Login to CloudBase
   

   AI will automatically open the login interface and guide environment selection

2. Start Developing

   Build a two-player online Gomoku game website, support online battle, and deploy it
   

   AI will automatically generate code, deploy to the cloud, and return the access link

We're not just "glue code". Our built-in rule library is designed specifically for AI programming, enabling AI to directly generate "deployable" CloudBase best-practice code.

Prompt: Generate a user login feature
- AI automatically generates code following CloudBase standards
- Auto-configure database, cloud functions, security rules
- One-click deploy to cloud

AI-automated MCP deployment flow, AI handles all cloud resource configuration from cloud functions, databases to static websites.

Prompt: Deploy current project to CloudBase
- Auto-detect project type (Web/Mini-program/Backend)
- Intelligent deployment parameter configuration
- Real-time deployment progress display
- Auto-return access link

Deployment error? Don't worry. AI automatically reads logs, analyzes issues, and generates fixes, truly achieving a develop-deploy-debug closed loop.

Prompt: There's an error: xxxx
- AI automatically views cloud function logs
- Analyze error causes
- Generate fix code
- Auto redeploy

Whether it's Web apps, mini-programs, or backend services, AI handles it all for you. You just focus on business logic.

Built-in intelligent vector search for CloudBase, WeChat Mini-Program and other professional knowledge bases, making AI understand CloudBase better.

Prompt: How to use cloud database to achieve real-time data synchronization?
- Intelligent search CloudBase knowledge base
- Return relevant documentation and best practices
- Provide code examples

Support /spec and /no_spec commands, intelligently choose based on task complexity.

/spec - Complete workflow (Requirements → Design → Tasks → Implementation)
/no_spec - Fast iteration (Direct implementation)

• Node.js v18.15.0 or above
• Enabled Tencent CloudBase Environment
• Installed AI IDE that supports MCP (View supported IDEs)

# Install
npm install @cloudbase/cli@latest -g

# Use
tcb ai

Add MCP configuration according to your AI IDE:

{
  "mcpServers": {
    "cloudbase": {
      "command": "npx",
      "args": ["@cloudbase/cloudbase-mcp@latest"]
    }
  }
}

{
  "mcpServers": {
    "cloudbase": {
      "command": "npx",
      "args": ["@cloudbase/cloudbase-mcp@latest"]
    }
  }
}

Development Process:

1. Input requirement: "Build a two-player online Gomoku website, support online battle"
2. AI generates: Web app + cloud database + real-time data push
3. Auto-deploy and get access link

Demo: Gomoku Game

View development screenshots

Development Process	Final Result

Development Process:

1. Input: "Develop a pet-raising mini-program with AI-enhanced interaction"
2. AI generates: Mini-program + cloud database + AI cloud function
3. Import to WeChat DevTools to publish

When an app has issues, AI automatically views logs, analyzes errors, and generates fix code.

39 tools covering environment management, database, cloud functions, static hosting, mini-program publishing, and other core features.

View complete tool documentation | Tool specification JSON

• Quick Start
• IDE Configuration Guide
• Project Templates
• Development Guide
• Plugin System
• FAQ

• Develop a Neighborhood Item Recycling Mini-Program with CloudBase AI CLI
• Build and auto-deploy with OpenClaw + CloudBase
• Agent Skills in practice: don't let AI-built code die on localhost
• One-stop development of card flip game with CodeBuddy IDE + CloudBase
• Develop a WeChat mini-game in 1 hour
• More tutorials...

• CloudBase: Use AI to develop an Overcooked game
• Software 3.0: Best AI Programming Partner
• More videos...

• Gomoku Online Game
• Overcooked Co-op Game
• E-commerce Admin Panel
• More cases...

Scan to join WeChat tech exchange group

Thanks to all the developers who have contributed to CloudBase MCP!