English
English
简体中文
日本語

AIFlow Tutorial

AIFlow User Guide

Introduction

AIFlow is a vibe coding tool designed specifically for M5Stack hardware and software. Without complex environment configuration, you can quickly generate code for M5Stack devices, deploy programs, and debug them using natural language.
AIFlow is available as both a desktop app and a mobile app. This guide uses connecting a StickS3 through the desktop app as an example to explain how to connect the device to AIFlow and quickly get started.

Devices Supported by AIFlow v1.0.2-beta

Device Supported
StickS3
StackChan
Cardputer-Adv

Software Downloads

Desktop App Download

PC Desktop Version Download Link
AIFlow-1.0.2-beta-setup.exe Download
AIFlow-1.0.2-beta-setup.zip Download
AIFlow-1.0.2-beta.dmg Download
AIFlow-1.0.2-beta.AppImage Download
AIFlow_1.0.2-beta_amd64.deb Download

Visit the AIFlow GitHub Releases page to download additional releases.

Mobile App Download

iOS users can search for AIFlow in the App Store to download the app. Android users can search for AIFlow on Google Play or download AIFlow-v1.0.1.apk.

Model Configuration

When you launch AIFlow for the first time, follow the getting-started guide to configure the model API:

The model API configuration primarily includes the following information:

  • Custom Name
  • Model ID
  • API Key
  • Model Provider Base URL
  • Test connection

After entering the information above, click Test connection. When Connection successful appears, the API has been configured successfully.

Currently supported models:

  • Anthropic Claude
  • Anthropic-compatible APIs provided by DeepSeek, Zhipu GLM, Alibaba Qwen, and others

Currently unsupported models:

  • OpenAI
  • Google Gemini

Flash Firmware

Before using AIFlow with a device, use the desktop app to flash the corresponding AIFlow firmware and configure the Wi-Fi information.

  1. Put the device into download mode: Connect the StickS3 to the computer, then press and hold the power button until the green indicator starts flashing. The device will enter download mode.
  1. After configuring the model API information in AIFlow, click Flash to open the firmware flashing page.
  1. Select the corresponding device port and firmware type, enter the Wi-Fi and other relevant information, then click Start Flash to begin flashing.

When Flash completed successfully appears, the AIFlow firmware has been flashed successfully.

Connect to AIFlow

After the firmware has been flashed, the device must establish a wireless connection with AIFlow using an Access Code. This connection is used for program deployment and device control.

  1. Obtain the device Access Code: After the StickS3 starts, press the center button once. When the server connection is established, the device will automatically generate an Access Code and display it on the screen.
  1. Enter the Access Code in AIFlow, then click Next to complete the device connection setup.

Getting Started

After completing the connection, enter natural-language instructions describing the functionality you want to implement. AIFlow displays the current task generation progress in real time. Once reasoning is complete, it automatically generates the corresponding MicroPython code and deploys it to the device over the network. Use the code preview and run-and-debug features to quickly validate and iterate on the result.

Note
Programs generated through AI inference are not deterministic. The same prompt may produce different program results. The following is provided only as a demonstration, and the actual result may vary.

Example

Print “Hello World” on screen

Result:

Interface Layout

The AIFlow interface is designed around the “chat -> code generation -> run and debug” workflow. It is arranged from left to right, making it easy to move quickly from an idea to a working result.

Common areas include:

  • Project Management
  • Guide and Theme Switching
  • Skills & MCP Extension Panel
  • Local Resource Management
  • Device File System
  • Application Generation Progress
  • Firmware Flashing & Device Management
  • Session Token Usage
  • Conversation History
  • Model Management
  • Token Usage History
  • Code Preview
  • Device Management
  • Device Remote Terminal

Skills & MCP

The Skills & MCP extension panel expands AIFlow's capabilities and is suitable for integrating external services beyond the basic programming workflow:

  • Import and manage custom Skills
  • SkillHub recommendations
  • MCP interface configuration and invocation

These capabilities can further extend AIFlow to scenarios such as weather queries, information retrieval, and smart home control.

File Management

The file system panel provides centralized management of local and device files:

  • Upload project assets, such as images and audio, to the local file list
  • Reference resource files directly in code
  • Send required resources and code to the device together during program deployment

This is suitable for applications that use UI assets or other local resources at runtime.

Token Usage

Token usage statistics for the current session are displayed in the upper-right corner of the Conversation area, making it easier to estimate model usage costs. In the upper-right corner of the Your Idea area, you can manage templates and view Token usage history for the past three days to help manage daily usage and quotas.

Mobile App

The overall workflow in the app is the same as in the desktop app:

  1. Configure the model API information.
  2. Bind the device using its Access Code.
  3. Enter your requirements in the chat box and deploy the program to the device.

More Information

The AIFlow project is continuously updated. Follow AIFlow GitHub for the latest updates.

Page Tools
PDF
On This Page