This is the third article in the MCP client getting-started tutorial series: "MCP Client Getting-Started Tutorial: ChatWise Installation and Configuration".

1. Introduction to ChatWise

ChatWise is a modern AI chat client, touted as "the second-fastest AI chatbot". It supports a variety of mainstream large language models such as ChatGPT, Claude, DeepSeek, Gemini, etc.

In addition to the basic AI conversation function, ChatWise also has the capabilities of web search, Artifacts generation, and MCP (Model Context Protocol) services. It is a powerful and user-friendly AI assistant tool.

The core advantages of ChatWise are:

• Lightweight: The program has a small size, starts quickly, and consumes few resources.
• Multi-model support: Integrates the APIs of various mainstream large AI models.
• Tool enhancement: Supports functions such as web search, file processing, and MCP services.
• Cross-platform: Supports Windows and macOS systems.
• Local deployment: High data security and can run completely locally.

Among them, the MCP service is a major highlight of ChatWise. By configuring MCP, ChatWise can obtain more powerful functions, such as file system access, web browsing, data processing, etc., greatly enhancing the practicality of the AI assistant.

2. Installation of ChatWise

1. Download ChatWise

First, we need to download and install the ChatWise client:

1. Open your browser and visit the ChatWise official website: https://chatwise.app/

2. Click the "Download" button on the homepage of the official website and select the corresponding version (Windows or macOS) according to your operating system.

3. After the download is complete, run the installation program and follow the wizard to complete the installation.

2. Initial Configuration

After the installation is complete, you need to perform basic configuration when running ChatWise for the first time:

1. Launch the ChatWise application.
2. Complete the initial settings according to the interface prompts (such as selecting the interface language, theme, etc.).

3. Configuration of AI model providers. If you want to use a specific AI model, you need to configure the corresponding API key in Settings > Providers.

ChatWise providers include: Anthropic, Azure OpenAI, DeepSeek, GitHub Copilot, Google AI, Groq, Mistral, Ollama, OpenAI, OpenRouter, Perplexity, Together, xAI, etc. You can apply for an API key from the corresponding provider and then fill it in.

Here, AIbase takes applying for an OpenRouter API key as an example to show you. Go to the API key application entry (OpenRouter https://openrouter.ai/settings/keys), select Create key, and name it to obtain the key.

After obtaining the key, copy it into the box and you're done.

4. Model selection. After configuring the provider, enter the conversation window and select the model you want in the upper left corner.

For example, AIbase selects the free model of DeepSeek here.

3. Prepare the MCP Environment

Before configuring the MCP service in ChatWise, we need to prepare the necessary environment:

1. Install Node.js

Most MCP services rely on the Node.js environment:

1. Visit the Node.js official website: https://nodejs.org/
2. Download and install the LTS (Long-Term Support) version.

3. After the installation is complete, open the terminal (command line) to verify the installation: Enter node -v. If the version number is displayed, it means the installation is successful. Enter npm -v. If the version number is displayed, it means the npm (Node Package Manager) installation is successful.

2. Install Python (Optional)

Some MCP services may require the Python environment:

1. Visit the Python official website: https://python.org/
2. Download and install the latest stable version.
3. After the installation is complete, open the terminal to verify the installation.

After preparing the basic environment, we can start configuring the MCP service in ChatWise.

4. Configure the MCP Service

ChatWise supports two types of MCP connection modes: stdio (Standard Input/Output) and SSE (Server-Sent Events).

This tutorial by AIbase will mainly introduce the SSE method.

1. Open the MCP Configuration Interface

1. Open the ChatWise application.
2. Click the avatar or user icon in the lower left corner.
3. Select "Settings" in the pop-up menu.
4. Select the "Tools" tab in the settings panel.

2. Add an MCP Service

On the tool settings page:

Method 1. SSE Mode

1. Click the plus (+) button in the lower left corner to add a new tool.
2. On the pop-up interface, select "Type" as "sse".

3. Add a service name (e.g., "fetch", named according to the service you want to add).

4. In the "url" input box, enter the SSE address startup command of the corresponding MCP service. For example, when AIbase enters the Magic Tower MCP Plaza, copy the address field and save it.

3. Save the Configuration

After completing the configuration, click the "Save" button in the upper right corner of the interface, or simply close the settings interface (ChatWise will automatically save the configuration).

Method 2. stdio Mode

1. Click the plus (+) button in the lower left corner to add a new tool.
2. On the pop-up interface, select "Type" as "stdio".
3. Add a service name (e.g., "fetch", named according to the service you want to add).
4. In the "Command" input box, enter the startup command for the corresponding MCP service parameters. For example, here I enter the following MCP service command for web browsing, and it will be automatically saved when you close the page.

npx   @playwright/mcp@latest

3. Configuration Examples of Common MCP Services

Here are some configuration command examples of common MCP services. You can selectively add them as needed:

File System MCP (Access Local Files)

npx -y @modelcontextprotocol/server-filesystem /path/to/Desktop /path/to/Downloads

Note: Please replace /path/to/Desktop and /path/to/Downloads with the actual paths on your computer.

• Windows example: npx -y @modelcontextprotocol/server-filesystem C:/Users/username/Desktop C:/Users/username/Downloads
• macOS example: npx -y @modelcontextprotocol/server-filesystem /Users/username/Desktop /Users/username/Downloads

Time Service MCP (Provides Time-related Functions)

uvx mcp-server-time --local-timezone Asia/Shanghai

Note: You can modify the time zone as needed. For example, the Beijing time zone is Asia/Shanghai.

Web Browsing MCP (Playwright)

npx @playwright/mcp@latest

This service allows the AI to interact with web pages, take screenshots, and execute JavaScript in a real browser environment.

Web Search MCP (Tavily)

npx -y tavily-mcp@0.1.4

Note: Using the Tavily search service requires an API key. Please add it to the environment variable: TAVILY_API_KEY="Your Key".

Sequential Thinking MCP (Improves Thinking Ability)

npx -y @modelcontextprotocol/server-sequential-thinking

This service can help the AI perform more structured thinking and reasoning.

4. Set Environment Variables (if necessary)

Some MCP services require an API key to work properly. In ChatWise, you can set environment variables like this:

1. On the tool configuration interface, find the MCP service you just added.
2. In the "Environment Variables" section, click Add.
3. Enter the name and value of the key in the format: Key Name="Key Value".

For example:

• Tavily search: TAVILY_API_KEY="your_api_key"
• AMap: AMAP_API_KEY="your_api_key"

5. Use the MCP Service

After the configuration is complete, we can use the MCP service in ChatWise:

1. Enable the MCP Service

1. Next to the conversation input box, click the "Tools" icon (usually a small hammer or toolbox icon).
2. Switch the toggle to the on state.
3. In the pop-up service list, select the MCP service you want to enable. Here, we only added the fetch and webview services, so you can see these two on the demonstration page.

2. Test the MCP Function

Now, you can ask ChatWise questions to test whether the MCP service is working properly. If you see information provided by the MCP service in the AI's response, it means the configuration is successful.

For example, after AIbase successfully completes the configuration above, just throw a URL and ask it to summarize the content in the link in 200 words and create a new title. The client will automatically call the fetch MCP server to complete the task.

6. Common Problems and Solutions

1. MCP Service Fails to Start

Problem Symptom: After adding an MCP service, you get a prompt saying "Unable to start the service" or there is no response when using it.

Possible Causes and Solutions:

• Incorrect command: Check if the command is entered correctly, especially the path part.
• Missing dependencies: Make sure you have installed the basic environment such as Node.js or Python.
• Permission issues: Try running ChatWise as an administrator.
• Path issues: Make sure there are no Chinese characters or special characters in the path.

2. API Key-related Problems

Problem Symptom: You have configured an MCP service, but you get a prompt saying "Invalid API key" or a similar error when using it.

Solutions:

• Check if the environment variables are set correctly.
• Confirm if the API key is valid (it may have expired).
• Check if the format of the environment variable is correct, usually in the form of: KEY_NAME="value".

3. Service Does Not Respond

Problem Symptom: The MCP service is enabled, but the AI assistant does not utilize the functions of this service.

Solutions:

• Confirm that the service is selected in the tool list.
• Restart the ChatWise application.
• Explicitly indicate that you need to use a specific function when asking a question (e.g., "Please use the file system to view my desktop").

8. Advanced Usage Tips

Combine Multiple MCP Services

You can enable multiple MCP services simultaneously to give the AI assistant more comprehensive capabilities. For example:

• Enable the page scraping service and the image generation service at the same time. The AI can handle page scraping content and generate images for the content.
• Combine web browsing and the file system. The AI can obtain information from the web and save it locally.

9. Summary

Through this tutorial, we have learned how to install the ChatWise client and configure the MCP service, and explored how to make the AI assistant more powerful. MCP technology allows AI to interact with the external world, greatly expanding the application scenarios of the AI assistant.

As the MCP ecosystem continues to develop, more powerful services will become available. It is recommended to regularly follow the updates of ChatWise and the dynamics of the MCP community to explore more possibilities.

We hope this tutorial will help you quickly get started with ChatWise and the MCP service, and make the AI assistant a more valuable tool in your work and life!

MCP Client Getting-Started Tutorial Series

《A Beginner's Guide to MCP for Novices: Learn to Use MCP Services from Scratch after Reading This》

《MCP Client Getting-Started Tutorial: Cherry Studio Installation and Configuration》

《MCP Client Getting-Started Tutorial: ChatWise Installation and Configuration》

《MCP Client Getting-Started Tutorial: Cursor Installation and Configuration》

《MCP Client Getting-Started Tutorial: trae Installation and Configuration》

《MCP Client Getting-Started Tutorial: Claude Installation and Configuration》