A local FastAPI web framework powered by Google Gemini for intelligent conversations and direct bash command execution.
It can answer and execute commands in succession until your task is completed.
- Rewritten from scratch: Now uses FastAPI (Backend) and HTML/CSS/JS (Frontend) for a robust and modern experience.
- Single Block Execution: No need for multiple terminals; running
main.pyhandles everything. - Premium UI: A brand new dark-themed, glassmorphism interface.
- Independent Chats: Each AI persona operates independently.
- Light/Dark Themes: Toggle between themes directly from the sidebar.
This is an AI chat framework designed for personal use,
transforming your local device into a powerful hub for AI-driven productivity and automation.
Built as a local web application with FastAPI,
it provides a seamless chat interface to interact with a sophisticated AI,
execute system commands, and manage the conversation history — all within your private environment.
I am not responsible for any code generated or executed by the AI (Google Gemini)
or provided by the user.
Use the assistant at your own risk.
Always review and validate any commands or scripts before running them.
I disclaim any liability for unintended consequences resulting from execution.
-
Local Web App Interface
Access your AI assistant through a modern, premium web interface hosted locally on your machine. Automatically opens in your browser upon startup. -
Google Gemini Integration
Leverage the power of Google's Gemini model for intelligent and context-aware conversations. -
Independent AI Chats
Each chat has its own memory and history and cannot access the memory of others. -
Direct Bash Command Execution (:)
Execute shell commands directly from the chat, enabling powerful system automation and fast task execution. -
Recursive Iterations (-r or -:r) The Bot can chat with itself using its output as his next prompt in order to complete your task step by step. It can also execute a bash command, read its output and continue with other commands until your task is completed.
-
Light/Dark Themes
Switch easily between light and dark modes. -
File-Based Chat History
Your conversation history is automatically saved to local text files.
To launch your personal AI assistant:
git clone https://github.com/gobbez/Skybound_AI
cd Skybound_AIpip install -r requirements.txtGet your API key from Google AI Studio.
Create a keys.py file in the root directory with:
GEMINI_KEY = "YOUR_API_KEY_HERE"python main.pyThe application will start the server and automatically open http://localhost:8000 in your default web browser.
These are the current commands that you can use to engage with the AI Assistant.
-
: Start your message with
:to execute a bash command (e.g ":ls", ":mkdir folder", ":nmap ip") -
-r Start your message with
-rto enable recursive chat. The bot will enter a loop passing its output as a next input in order to answer your task. It will automatically stop whenever it thinks the task is answered. (e.g "-rWrite numbers from 1 to 5. You can write one number per message") -
-:r Start your message with
-:rto enable recursive commands. The bot will enter a loop executing bash commands and using the command output as input for his successive command(s). It will automatically stop whenever it thinks the task is completed. (e.g "-:rStep1: check my ip. Step2: ping my ip. Execute one step per message") -
-f Start your message with
-ffor file creation. The bot will process and create a file with the content you provide. Format:-f filename.ext: content
Anyone is free to contribute.
This project embraces AI, Bash, Penetration Testing and modularity, so everyone can join!
This project is open-source and available under the MIT License
Hi, I'm Andrea! I'm a Python backend developer. Recently, I dove into the world of AI, Data Science and Penetration Testing. I’m passionate about creating useful bots and experimenting with automation.
This project is a personal playground combining my curiosity for AI, Bash, and productivity tools, with a hint of Penetration Testing too!