aiexeauto-gui is an autonomous AI agent that thinks and acts on its own. This innovative application understands natural language input and performs computer tasks in the optimal way, automatically handling even complex tasks.
Watch our demo video on YouTube to get a quick overview of what aiexe can do for you! Click here to watch the video.
- Safe Execution: Operates securely based on Docker.
- Local LLM Support: Uses locally running LLMs to reduce costs and ensure data security.
- Natural Language Input: Executes tasks with simple commands like "Organize the files" or "Edit the video."
- Automation Features: Automates various repetitive tasks such as file management, image editing, and document processing.
- Cross-Platform Support: Supports both Windows and macOS.
- Real-Time Progress Display: Check task progress through a clean interface.
- MCP Support: MCP (Model Context Protocol) is supported and runs inside the Docker container when Docker is used.
- Operating System: Windows, macOS
-
Download and install Node.js from nodejs.org.
-
Download and install Python from python.org.
For general users, AIEXEAUTO-GUI can be installed easily by following the steps below.
- Run PowerShell as Administrator and execute the following command:
Set-ExecutionPolicy RemoteSigned -Scope CurrentUser -Force; New-Item -ItemType Directory -Path "~/.aiexeauto" -Force | Out-Null; Get-ChildItem Env: | ForEach-Object { "$($_.Name)=$($_.Value)" } > ~/.aiexeauto/.env; [System.IO.File]::WriteAllLines("$HOME\.aiexeauto\.env", (Get-ChildItem Env: | ForEach-Object { "$($_.Name)=$($_.Value)" }), [System.Text.UTF8Encoding]::new($false)); if (Get-Command npm -ErrorAction SilentlyContinue) { $timestamp = Get-Date -Format "yyyyMMddHHmmss"; $folderName = "_aiexeauto-gui_project_$timestamp"; $desktopPath = [System.IO.Path]::Combine([System.Environment]::GetFolderPath('Desktop'), $folderName); New-Item -ItemType Directory -Path $desktopPath -Force; if (Test-Path $desktopPath) { Set-Location -Path $desktopPath; Invoke-WebRequest -Uri "https://github.com/kstost/aiexeauto-gui/archive/refs/heads/main.zip" -OutFile "__aiexeauto-gui_project__.zip" -ErrorAction Stop; if (Test-Path "__aiexeauto-gui_project__.zip") { Expand-Archive -Path "__aiexeauto-gui_project__.zip" -DestinationPath "."; Set-Location -Path "aiexeauto-gui-main"; npm i; if ($?) { npm run build; if ($?) { ii "dist\\aiexeauto Setup*.exe" } } } } } else { Write-Output "npm is not installed. Please download and install it from https://nodejs.org." }
- Run Terminal and execute the following command:
sudo chown -R 501:20 ~/.npm 2>/dev/null; mkdir -p ~/.aiexeauto; env > ~/.aiexeauto/.env; command -v npm >/dev/null 2>&1 && { timestamp=$(date +%Y%m%d%H%M%S) && cd ~/Downloads && mkdir "_aiexeauto-gui_project_$timestamp" && cd "_aiexeauto-gui_project_$timestamp" && git clone https://github.com/kstost/aiexeauto-gui && cd aiexeauto-gui && npm i && npm run build && open dist/aiexeauto-*.dmg; } || { echo "npm is not installed. Please download and install it from https://nodejs.org."; }
Docker is optional. Using Docker allows you to perform tasks in a safer and isolated environment. If you want to use Docker, please follow these steps.
- All operations will affect your actual system files and settings
- There is no isolation between the application and your system
- We strongly recommend against using automatic code execution mode without Docker
- Always review generated code carefully before execution
- Consider backing up important data before running any tasks
-
Install Docker
- Download and install the latest version of Docker Desktop.
-
Build the Docker Image
For macOS:
git clone https://github.com/kstost/aiexeauto-gui.git && cd aiexeauto-gui/my-docker-app && docker build --platform linux/x86_64 -t my-node-ubuntu .
For Windows:
- Open "PowerShell" as Administrator from the Start menu.
- Execute the following commands sequentially:
# Change execution policy Set-ExecutionPolicy -Scope CurrentUser -ExecutionPolicy RemoteSigned -Force # Create directory New-Item -ItemType Directory -Path "my-docker-app" -Force # Navigate to the directory cd my-docker-app # Download Dockerfile Invoke-WebRequest -Uri "https://raw.githubusercontent.com/kstost/aiexeauto-gui/refs/heads/main/my-docker-app/Dockerfile" -OutFile "Dockerfile" # Build Docker image docker build --platform linux/x86_64 -t my-node-ubuntu .
Ollama is a locally running LLM. If you wish to use Ollama, follow the steps below for installation.
-
Install Ollama
- Download and install the latest version of Ollama.
-
Download the Ollama Model
ollama run qwen2.5
Model exploration is available at Ollama Search. Since AIEXEAUTO-GUI uses tool-based models, please select models under the "Tools" category.
Once setup is complete, select the folder containing the files you want to process and enter the desired task description in natural language. The AI will automatically perform the task.
During the process, the AI generates and executes necessary code through multiple steps. Users can review the generated code before execution.
When all tasks are completed, the folder containing the results will au 7BB6 tomatically open. Additionally, the opened folder will be set as the default folder for the next task, allowing users to continue without reselecting folders.
However, the context of the previous task is not shared with the next task.
- Data Backup: Always back up important data before executing tasks.
- Security: Be cautious about security when allowing the application to perform real-world tasks.
- API Costs: Tasks may incur API fees for services like Claude, DeepSeek, OpenAI, Groq, and Gemini. Currently, only Gemini and OpenAI are temporarily supported as API providers.
- macOS Docker Usage: When using Docker on macOS, please note that the application directly modifies data in the connected folder for data processing.
- Common Errors: Check for API keys, paths, and permission issues.
- Support Request: If additional support is needed, visit COKAC website or the "코드깎는노인" community.
This software is distributed under the MIT License. For details, refer to the LICENSE file.
- Bug reports, feature requests, and pull requests are welcome.
- Please include test code when contributing.
This software is in the prototype stage. Users are fully responsible for its use, and caution is advised when using it with important data or systems.