# CBK AI MCP Server

> Model Context Protocol (MCP) server for CBK AI (ChatBotKit) - providing seamless access to documentation, API references, and integration guides through MCP-compatible clients.

The CBK AI MCP server enables AI development tools like Claude Desktop, VS Code, Zed, and other MCP-compatible clients to access comprehensive ChatBotKit documentation, perform semantic searches across manuals and guides, and retrieve up-to-date API references during development.

Key features:
- Documentation and manual search capabilities
- API reference lookup
- Code examples and integration patterns
- API operations for resource management (with authentication)
- Bootstrap resources, query status, and manage platform entities
- Supports all MCP-compatible clients

## Model Context Protocol (MCP) Server

CBK AI provides a Model Context Protocol server for seamless integration with MCP-compatible clients like Claude Desktop, Zed, VS Code, and other AI development tools.

### Claude Desktop

```json
{
  "mcpServers": {
    "cbk": {
      "command": "npx",
      "args": [
        "-y",
        "@modelcontextprotocol/server-fetch",
        "https://mcp.cbk.ai/mcp"
      ],
      "env": {
        "CBK_SECRET": "your-api-secret-here"
      }
    }
  }
}
```

Replace `your-api-secret-here` with your ChatBotKit API secret.

Note: The API secret is only required for interacting with ChatBotKit resources and objects. Querying content such as docs, manuals, and tutorials does not require authentication.

### Claude Code

Use the Claude CLI to add the CBK MCP server:
```bash
claude mcp add --transport http cbk https://mcp.cbk.ai/mcp \\
  --header "Authorization: Bearer your-api-secret-here"
```

Alternatively, within Claude Code, type `/mcp` to authenticate and manage MCP servers.

Note: The Authorization header is only required for interacting with ChatBotKit resources and objects. Querying content such as docs, manuals, and tutorials does not require authentication.

### OpenAI Codex

To configure CBK MCP for OpenAI Codex, use the CLI:
```bash
codex mcp add cbk -- npx -y @modelcontextprotocol/server-fetch https://mcp.cbk.ai/mcp
```

To add authentication, edit your `~/.codex/config.toml` file and add the Authorization header:
```toml
[mcp_servers.cbk]
command = "npx"
args = ["-y", "@modelcontextprotocol/server-fetch", "https://mcp.cbk.ai/mcp"]

[mcp_servers.cbk.env]
CBK_SECRET = "your-api-secret-here"
```

Note: The API secret is only required for interacting with ChatBotKit resources and objects. Querying content such as docs, manuals, and tutorials does not require authentication.

### Cursor

1. Open Cursor Settings (Cmd/Ctrl + ,)
2. Navigate to "Features" → "MCP" or search for "MCP"
3. Click "Add New MCP Server"
4. Configure the server with these settings:
```
Server Name: cbk
Type: http
URL: https://mcp.cbk.ai/mcp
Headers:
  Authorization: Bearer your-api-secret-here
```

Alternatively, edit the `~/.cursor/mcp.json` file directly:
```json
{
  "servers": {
    "cbk": {
      "type": "http",
      "url": "https://mcp.cbk.ai/mcp",
      "headers": {
        "Authorization": "Bearer your-api-secret-here"
      }
    }
  }
}
```

Note: The Authorization header is only required for interacting with ChatBotKit resources and objects. Querying content such as docs, manuals, and tutorials does not require authentication.

### GitHub Copilot

1. Open VS Code Command Palette (Cmd/Ctrl + Shift + P)
2. Run "MCP: Add Server" command
3. Configure the server:
```
Server Name: cbk
Type: http
URL: https://mcp.cbk.ai/mcp
Headers:
  Authorization: Bearer your-api-secret-here
```

Alternatively, create/edit `.vscode/mcp.json` in your workspace:
```json
{
  "servers": {
    "cbk": {
      "type": "http",
      "url": "https://mcp.cbk.ai/mcp",
      "headers": {
        "Authorization": "Bearer your-api-secret-here"
      }
    }
  }
}
```

Note: The Authorization header is only required for interacting with ChatBotKit resources and objects. Querying content such as docs, manuals, and tutorials does not require authentication.

### GitHub Copilot Agent

To configure CBK MCP for GitHub Copilot Agent in your repository:
1. Navigate to your repository on GitHub
2. Go to Settings → Code & automation → Copilot → Coding agent
3. Add the following MCP configuration in the MCP configuration section:
```json
{
  "mcpServers": {
    "cbk": {
      "type": "http",
      "url": "https://mcp.cbk.ai/mcp",
      "headers": {
        "Authorization": "$COPILOT_MCP_CBK_SECRET"
      },
      "tools": ["*"]
    }
  }
}
```

Then, add your API secret to the Copilot environment:
4. Go to Settings → Environments
5. Create or select the `copilot` environment
6. Add a secret named `COPILOT_MCP_CBK_SECRET` with your API secret

Note: The Authorization header is only required for interacting with ChatBotKit resources and objects. Querying content such as docs, manuals, and tutorials does not require authentication.

### Gemini Code Assist

1. Ensure you have the Gemini CLI installed and configured
2. Create or edit your MCP configuration file
3. Add the CBK MCP server configuration:
```json
{
  "mcpServers": {
    "cbk": {
      "type": "http",
      "url": "https://mcp.cbk.ai/mcp",
      "headers": {
        "Authorization": "Bearer your-api-secret-here"
      }
    }
  }
}
```

Note: Gemini Code Assist MCP support is still evolving. Check the official documentation for the latest setup instructions and supported features. The Authorization header is only required for interacting with ChatBotKit resources and objects. Querying content such as docs, manuals, and tutorials does not require authentication.

## What the MCP Server Provides

**Without Authentication (Public Access):**
- Documentation and manual search capabilities
- API reference lookup
- Code examples and integration patterns
- Real-time access to the latest CBK AI features and updates

**With Authentication (Authenticated Access):**
- Execute API operations against the CBK AI platform
- Bootstrap and manage resources (bots, datasets, skillsets, conversations)
- Query resource status and configurations
- Full access to platform functionality
- Create, update, delete, and manage platform entities

## Authentication

The MCP server is publicly accessible without authentication for documentation access. To enable API operations and resource management, configure authentication using your CBK AI API secret.

To obtain your API secret:
1. Sign in to [ChatBotKit Console](https://console.cbk.ai)
2. Navigate to Settings → API Keys
3. Create a new API key or use an existing one

**Important:** The Authorization header (API secret) is only required for interacting with ChatBotKit resources and objects (bots, datasets, conversations, etc.). Querying content such as docs, manuals, and tutorials does not require authentication.

**Privacy & Security:**
- API secrets are transmitted securely over HTTPS
- Secrets are never logged or stored by the MCP server
- No data used for service improvements
- Your data and credentials remain private

## Optional

- [CBK AI Platform](https://cbk.ai): Main platform for production deployments
- [Developer Documentation](https://docs.cbk.ai): Complete SDK and API documentation (see https://docs.cbk.ai/llms.txt)
- [Discord Community](https://go.cbk.ai/discord): Connect with other developers