The AI Agent is an intelligent desktop automation tool designed as a goal-oriented agent capable of handling a wide range of tasks, from file management to document creation and even application control. It acts as your digital co-pilot, automating repetitive work, securely managing files, and streamlining day-to-day operations. Leveraging Agentic AI and Large Language Models (LLMs), this agent understands and executes complex instructions, bridging the gap between natural language commands and system actions.
This AI Agent also accepts instructions via speech, enabling hands-free operation of your computer. This showcases an early form of Embodied AI Agents within a desktop environment, allowing for more natural Human-Agent Collaboration.
- Plan Execution: Create, copy, move, rename files/folders based on intricate patterns, demonstrating Agentic AI Planning and Reasoning.
- Information Retrieval: Search files and folders efficiently, a fundamental aspect of agent information gathering.
- Action Execution: Open files with their default applications, a basic tool utilization for agents.
- State Awareness: List files and folders in a specific directory, crucial for an agent to understand its environment.
- Secure Operations: Encrypt files using secure algorithms, demonstrating Agent Ethics and Safety by maintaining data confidentiality.
- Integrity Assurance: Maintain confidentiality and integrity of sensitive data.
- Tool Integration: This is accomplished using the
cryptographylibrary, highlighting Tools and Frameworks for Agentic AI.
- Content Creation Automation: Generate
.docxfiles with advanced formatting,.pptxpresentations with custom styles, and convert documents to PDF. This exemplifies Agentic AI Applications in content creation. - LLM-Powered Generation: This feature is enhanced by Agentic AI and Large Language Models (LLMs), enabling the agent to understand and generate content based on complex prompts.
- Environment Interaction: Open installed programs and control them, execute CLI commands directly through the agent. This showcases an agent's ability to interact with its digital environment.
- Workflow Automation: Automate workflows in terminal/command prompt style, demonstrating Agentic AI for Complex Problem Solving.
- Environmental Access: Open specific folders instantly, launch applications on demand.
- Window Control: Control program windows (minimize, maximize, close), further enhancing the agent's interaction with the desktop environment.
git clone https://github.com/Anish-CodeDev/System_AI_Agent/
cd System_AI_Agentpython -m venv venv
source venv/bin/activate # Mac/Linux
venv\Scripts\activate # Windowspip install -r requirements.txtThe AI Agent's backend relies on the system_agent mcp server. For insights into its underlying architecture and the development of such Agentic AI Architectures, explore: https://github.com/Anish-CodeDev/mcp_servers/
The generated_documents folder showcases outputs from our agent's generative capabilities. The encrypted_documents folder demonstrates its security-focused operations.
python app.pyThese examples illustrate the agent's ability to interpret and execute instructions, a key aspect of Agentic AI Planning and Reasoning.
> create file report.txt
> encrypt file secrets.docx
> generate presentation "Applications of AI" named presentation.pptx
> open folder Documents
> generate a word document on the topic "World Models" named world_model.docx
> open program "Notepad", copy all the text
- Constrained Actions: The AI Agent is designed with safety constraints; it does not possess the authority to delete any file, focusing only on creation, copying, and moving. This reflects a commitment to Agent Ethics and Safety.
- Authorization Protocols: The AI Agent only executes commands you explicitly authorize, ensuring user control and preventing unintended actions, a crucial aspect of Human-Agent Collaboration.
- Core System: Python 3.9+
- Agentic Libraries:
load_dotenv- For secure loading of API Keys.cryptography– For robust Encryption/Decryption, underpinning agent security.python-docx– For Word document generation, an Agentic AI Application.docx2pdf- For converting documents to PDF, enhancing agent output usability.pandas- For data manipulation, supporting complex agent tasks.python-pptx– For PowerPoint creation, another Agentic AI Application.pyautogui– For GUI automation, enabling the agent to interact with the visual interface.subprocess– For CLI execution, a fundamental Tool for Agentic AI.os,shutil– For file operations, core to an agent's environmental interaction.google-genai- Provides access to Gemini 2.5 LLM's, powering Agentic AI and Large Language Models (LLMs).langgraph- Manages the agentic workflow, crucial for Agentic AI Architectures and complex reasoning.mcp- Provides an interface for Agent Communication and Coordination between client and server components.pyside6- Provides a GUI interface for user interaction, facilitating Human-Agent Collaboration.
This project is licensed under the MIT License – you are free to use, modify, and distribute it, fostering open development in Tools and Frameworks for Agentic AI.
- Advanced Reasoning & Planning: Enhance AI Agent Planning and Reasoning for more complex, multi-step tasks.
- Voice Command Integration: Further develop Embodied AI Agents by refining voice command support for seamless interaction.
- Cloud Storage Integration: Expand agent capabilities to interact with cloud-based data stores.
- AI-Powered Task Suggestions: Implement Self-Improving AI Agents that can suggest tasks based on user patterns and context.
- Real-time Activity Monitoring: Develop sophisticated Agent Ethics and Safety features through real-time monitoring.
- Multi-Agent System Development: Explore the potential for Multi-Agent Systems, enabling agents to collaborate on tasks.
- Enhanced Learning Capabilities: Integrate Learning in Agentic AI to adapt and improve performance over time.
- Robust Evaluation Metrics: Define and implement Agentic AI Evaluation Metrics to quantify agent performance and progress.