AI-Assisted Development

AI Chatbot

The Ameba AIoT AI Chatbot is an intelligent assistant provided by Kapa AI. It is designed to provide developers with efficient and accurate technical support and documentation query services. By conversing with the Ameba AIoT AI Chatbot, users can quickly obtain relevant information from Ameba’s official documentation and resolve issues encountered during development.

Key Features

• Intelligent Documentation Q&A: Provides accurate technical question answers based on the Ameba official documentation library

• Code Example Generation: Generates corresponding code examples and configuration instructions based on user requirements

• Problem Diagnosis Assistance: Helps developers locate and resolve issues encountered during development

• Multi-language Support: Supports Chinese and English documentation queries

Knowledge Sources

The Ameba AIoT AI Chatbot knowledge base covers the following content:

• Chip Products: Technical specifications, performance parameters, and power consumption characteristics of the complete Ameba chip series including RTL8721Dx, RTL8720E, RTL8710E, RTL8726E, RTL8730E, etc.

• Datasheets: Chip electrical characteristics, pin definitions, storage specifications, peripheral interfaces, etc.

• SDK Documentation: API references, development guides, and driver descriptions for development platforms including FreeRTOS, Linux, Zephyr, Arduino, and Matter

• Key Technologies: Core technical documentation for Wi-Fi R-MESH, Wi-Fi 6, Bluetooth LE 5.x, AT Commands, Secure Boot, TrustZone, etc.

• GitHub: ameba-rtos XDK code examples

Access Methods

Users can access the Ameba AIoT AI Chatbot through the following method:

Web Access: Open RealMCU AIOT, click the AI icon in the bottom right corner to start a conversation with the chatbot.

Documentation MCP Service Guide

Overview

MCP (Model Context Protocol) is an open standard protocol introduced by Anthropic to help AI Agents securely connect with external tools, data sources, and workflows. Through MCP, AI Agents can dynamically call external services and expand their capabilities.

The Ameba AIoT documentation provides an MCP server, sharing the same knowledge source with the AI Chatbot. After installing the Ameba documentation MCP server, AI Agents (such as Claude Code, Cline, Continue, Roo Code) can directly query and retrieve Ameba AIoT official documentation content.

Compared to the AI Chatbot, the documentation MCP server can:

• Integrated into Development Environment: Through the MCP protocol, AI Agents can directly obtain documentation content and modify project code in the editor without interacting through a chat interface

• Provide More Structured Query Results: The documentation content returned by the MCP server is processed and formatted, making it more suitable for AI Agents to parse and use

• Wider AI Agent Support: Supports various AI Agents (such as Claude Code, Cline, Continue, Roo Code, etc.), meeting the needs of different developers

Note

To learn more about MCP protocol detailed specifications, please visit: MCP Official Documentation

MCP Server Capabilities

The Ameba documentation MCP server is based on HTTP protocol and provides the following tools for AI Agents:

MCP Tools List

Tool	Description
search_realtek_knowledge_sources(query)	Performs semantic search on Realtek documentation and other knowledge bases, returning the most relevant text fragments and source links, providing AI Agents with official reference materials.

MCP Service Configuration

To obtain the MCP server address, follow these steps:

1. Open the RealMCU AIOT webpage

2. Click the AI icon in the bottom right corner to open the RealMCU AI Chatbot

3. Find the Use MCP dropdown in the pop-up dialog

4. For Claude Code, click “Add to Claude Code” and copy the CLI command

5. For other AI Agents like Roo Code, Cline, Continue, click “Copy MCP Link” to get the MCP server address

Note

MCP Server Address Format:

https://ameba-aiot.mcp.kapa.ai

After obtaining the MCP service address, configure your AI Agent according to the following sections.

Access Steps

The following are the general steps to access the Ameba documentation MCP service:

1. Obtain MCP Service Address: Get it from the RealMCU AIOT webpage (see MCP Service Configuration)

2. Configure MCP Client: According to the AI programming assistant you use, configure according to the corresponding section below

3. Complete Authentication: Execute authentication in the AI Agent

4. Start Using: After successful authentication, you can query documentation content

Claude Code Configuration

Step 1: Add MCP Service

Run claude in the terminal to start Claude Code, and execute the following command to add the Ameba documentation MCP service:

claude mcp add --transport http realmcu-ask-ai-docs https://ameba-aiot.mcp.kapa.ai

Parameter description:

• --transport http: Use HTTP transport protocol

• realmcu-ask-ai-docs: MCP server name (can be customized)

• https://ameba-aiot.mcp.kapa.ai: MCP service address

Step 2: Complete Authentication

1. Restart Claude Code to apply the configuration

2. In the Claude conversation, enter:

   Help me with ameba doc MCP authentication
   

3. Follow Claude’s prompts to complete the authentication process

Tip

• View added MCP list: claude mcp list

• Remove MCP service: claude mcp remove realmcu-ask-ai-docs

Cline Configuration

Step 1: Add MCP Service

Add through the VS Code settings interface:

1. Go to Cline -> MCP Servers -> Remote Servers and fill the following content:

   Name: realmcu-ask-ai-docs
   URL: https://ameba-aiot.mcp.kapa.ai
   Transport Type: Streamable HTTP
   

2. Click the Add Server button to add the Ameba documentation MCP service

Step 2: Complete Authentication

1. Go to the Configure interface, find the realmcu-ask-ai-docs MCP service just added, click the Authenticate button

2. Follow the prompts to complete the authentication process

Roo Code Configuration

Step 1: Add MCP Service

Add the Ameba documentation MCP service in the Roo Code MCP configuration file:

1. Open the Roo Code Settings interface

2. Find the MCP Servers configuration section, edit the JSON configuration file

   • Click Edit Global MCP to open the global MCP configuration file mcp_config.json

   • Click Edit Project MCP to open the current workspace MCP configuration file .roo/mcp.json

Add the following content:

{
   "mcpServers": {
      "realmcu-ask-ai-docs": {
         "type": "streamable-http",
         "url": "https://ameba-aiot.mcp.kapa.ai",
         "disabled": false,
         "alwaysAllow": [],
         "headers": {
            "Authorization": "Bearer <your_personal_access_token>"
         }
      }
   }
}

Roo Code currently has incomplete support for Remote MCP Server OAuth 2.0 authentication flow and cannot automatically pop up third-party OAuth (such as GitHub) authentication pages. It is recommended to use MCP Inspector to complete the authentication, extract the Token, and fill it into the Roo Code configuration as a temporary solution.

Step 2: Complete Authentication

Use the official MCP Inspector tool to complete GitHub OAuth login, obtain the Access Token, and fill it into the Roo Code configuration.

1. Install and start MCP Inspector

   npx @modelcontextprotocol/inspector
   

2. Configure the connection in Inspector

   Transport Type: Streamable HTTP
   URL: https://ameba-aiot.mcp.kapa.ai
   

   Click Connect → The GitHub OAuth authentication page will pop up normally → Complete authorization

3. Extract Token from browser developer tools

   After completing GitHub authorization, open browser F12 → Network panel, find the request to ameba-aiot.mcp.kapa.ai, check the request header: Authorization: Bearer eyJhbGxxxxxx…, copy the Token value.

4. Fill the Token into Roo Code configuration

   {
      "mcpServers": {
         "realmcu-ask-ai-docs": {
            "type": "streamable-http",
            "url": "https://ameba-aiot.mcp.kapa.ai",
            "disabled": false,
            "alwaysAllow": [],
            "headers": {
               "Authorization": "Bearer eyJhbGxxxxxx..."
            }
         }
      }
   }
   

5. The realmcu-ask-ai-docs MCP service can now be used in Roo Code.

Continue Configuration

Step 1: Add MCP Service

Add through the VS Code settings interface:

1. Go to Continue -> Open Settings -> Tools, click the + button of MCP Servers

2. In the opened new-mcp-server.yaml, replace the following content to add the Ameba documentation MCP service:

   name: My Continue Config
   version: 0.0.1
   schema: v1
   mcpServers:
   - name: realmcu-ask-ai-docs
     type: streamable-http
     url: https://ameba-aiot.mcp.kapa.ai
     requestOptions:
         headers:
         Authorization: "Bearer eyJhbGxxxxxx..."
   

It is recommended to use MCP Inspector to complete the authentication, extract the Token, and fill it into the Continue configuration.

Step 2: Complete Authentication

Use the official MCP Inspector tool to complete GitHub OAuth login, obtain the Access Token, and fill it into the Continue configuration. 1. The steps to obtain the Token are the same as steps 1-3 in Roo Code Configuration Authentication

2. Fill the Token into the Continue configuration

   name: My Continue Config
   version: 0.0.1
   schema: v1
   mcpServers:
   
   - name: realmcu-ask-ai-docs
   
     type: streamable-http
     url: https://ameba-aiot.mcp.kapa.ai
     requestOptions:
         headers:
         Authorization: "Bearer <your_personal_access_token>"
   

5. The realmcu-ask-ai-docs MCP service can now be used in Continue.

Query Examples

Taking Claude Code as an example, here are actual examples of querying Ameba documentation:

Example 1: Query OTA Feature Overview

Example 2: Query GPIO Usage

FAQ

Q1: What should I do if the MCP service connection fails?

A: Please check the following:

• Confirm that the network can access ameba-aiot.mcp.kapa.ai

• Confirm that the authentication Token is correct

Q2: Can I continue to use it after authentication?

A: The authentication validity depends on the Token type, usually valid for a long time. To re-authenticate, tell the AI Agent “Re-authenticate MCP” or “Perform MCP authentication again”.

Q3: Can I configure multiple MCP services at the same time?

A: Yes, you can add multiple MCP servers in the same configuration file, just use different names.

Q4: What if the AI Agent does not actively call the MCP service?

A: It may be because the current conversation context does not trigger the conditions for calling MCP. Try explicitly telling the AI Agent to use MCP to query related information, for example, “Please query Ameba’s OTA functionality through the realmcu-ask-ai-docs MCP service”.