ForgeOAgent

Intelligent automation for everyone:
Use Gemini and custom agents to improve prompts, enhance text, refine code, generate emails, and automate tasks.

ForgeOAgent is a modular Python framework for building, managing, and running AI agents with advanced code processing capabilities. It automates code improvement, agent management, and client integration using Google Gemini and other APIs.

🚀 Key Features

• Cross-platform shell support: Works on Windows & Linux.
• Flexible agent modes:
  • Inquirer Mode: Use system prompts for quick tasks.
  • Executor Mode: Advanced workflows, agent management, and context-aware automation.
• Customizable prompts: Easily add your own system instructions for new agent behaviors in mcp/system_prompts/
• Persistent chat history: Optionally load recent conversations for better context and accuracy.
• Automatic tool integration: Place Python classes in mcp/tools to auto-expose new agent capabilities.
• Comprehensive logging: All agent activities and improvements are logged for review.

📁 Project Structure

• start.py — GUI entry point for the agent system.
• main.py — Command-line entry point for running agents and workflows.
• agents/ — Agent modules and metadata.
• clients/ — API clients for external services (e.g., Gemini).
• core/ — Core utilities for agent management, API keys, configuration, and more.
• logs/ — JSONL logs of agent activities and improvements.
• shell/ — Shell scripts for Linux and Windows environments.
• mcp/ — System prompts, agent context, and tools for modular extension.

UseFul :

• GeminiAPIClient().generate_content
• GeminiAPIClient().search_content
• from forgeoagent.controller.executor_controller import create_master_executor

use :

• from forgeoagent.controller.executor_controller import print_available_executors , save_last_executor
• from forgeoagent.controller.inquirer_controller import inquirer_using_selected_system_instructions , print_available_inquirers , auto_import_inquirers

🛠️ Getting Started

1. Clone the repository

   git clone https://github.com/ANGELKORADIYA/ForgeOAgent.git
   cd ForgeOAgent
   

2. Install dependencies

   pip install -r requirements.txt
   

3. Configure API keys

   • Copy .env.production to .env and add your Gemini API keys.

4. Run the agent system

   python main.py
   

   • List available system prompts:
     python main.py -l
   • Run with a specific prompt:
     python main.py -p <prompt_type> "your prompt here"
   • Use main mode (with agent context):
     python main.py "your prompt here" --main

5. Assign a shortcut for quick access

   • Linux:
     Add a keyboard shortcut to run shell/linux/start.sh.
   • Windows:
     Create a desktop shortcut to shell/windows/start.vbs.

💡 Usage Examples

• Quick prompt improvement:
  Use Inquirer Mode to enhance text, code, or emails with a single command.
• Context-aware automation:
  Use Executor Mode to leverage previous conversations and agent context for more complex tasks.

⚙️ Customization

• Add new system prompts:
  Place your custom prompt as a Python constant in mcp/system_prompts/ (naming: *_SYSTEM_INSTRUCTION).
• Integrate new clients:
  Add API clients in clients/.
• Review logs:
  Check logs/ for detailed agent activity and improvements.
• Main Agent Reuse: Save Result

🧩 Creating Custom Agents & Shortcuts

1. Add your prompt:
   Create a new file in mcp/system_prompts/ and define a constant ending with _SYSTEM_INSTRUCTION.
2. Create a shortcut script:
   Use .sh (Linux) or .vbs (Windows) to call the agent with your prompt type.

Now you can trigger your custom agent instantly via your shortcut.

📝 Example: Improving Code Files

• Recursively process files in a directory (skipping .git).
• Use Gemini agents to improve code or text based on file extension.
• Improved files are saved in the current directory.
• See agents/temp_3/main_agent.jsonl for a sample agent log and code.

🤝 Contributing

Contributions are welcome! Please submit issues or pull requests for improvements or new features.

📄 License

MIT License