| AutoGen Studio 0.4 Custom Agents

Unlock unparalleled automation and human-AI collaboration with a curated suite of specialized agents, tools, and configurations for AutoGen Studio 0.4. This collection extends default capabilities with powerful components for enhanced human-in-loop conversations and complex task automation.

---

Features

• Diverse Agent Types: Includes specialized agents like virtual assistants, chatbots, configuration experts, and more.
• Extended Model Support: Configurations for various large language models including GPT-4o mini, Llama3.1, DeepSeek Coder, and others.
• Comprehensive Toolset: Over 15 robust tools ranging from calculators to API integration and workflow automation, empowering agents to interact with the real world.
• Flexible Termination Conditions: Multiple termination strategies for fine-grained conversation control and efficient task completion.
• Human-in-Loop Teams: Pre-configured teams for seamless collaborative human-AI interaction, ensuring oversight and intelligent intervention.

---

Components Overview

Agents

• human_in_loop_team: A pre-configured team designed for collaborative human-AI interaction, allowing for guided workflows and critical decision points.
• llama3_agent: An versatile assistant powered by Ollama's Llama3.1 model, ideal for general conversational tasks.
• virtual_assistant: Specialized for booking appointments, making reservations, and managing schedules.
• chatbot: A natural language conversational agent for engaging and responsive dialogue.
• config_expert: An agent focused on validating and modifying configurations for agents, models, and tools.
• model_config: Manages and optimizes various model configurations, ensuring agents use the right intelligence for the task.
• tool_config: Manages and optimizes tool configurations, ensuring efficient and accurate tool execution.

Models

• GPT-4o mini
• Llama3.1
• DeepSeek Coder v2
• Qwen2.5 Coder
• Dolphin-Llama3
• CodeLlama 13b

Tools

• Basic Tools:
  • calculator: Performs mathematical operations.
  • website_fetcher: Retrieves content from web pages.
• Advanced Tools:
  • task_parser: Breaks down complex prompts into manageable sub-tasks.
  • domain_knowledge_base: Accesses and queries specialized knowledge sources.
  • model_trainer: Facilitates training or fine-tuning of machine learning models.
  • workflow_automation: Automates multi-step processes and orchestrates agent actions.
  • api_integration: Connects to external APIs for enhanced capabilities.
  • error_handling: Provides mechanisms for identifying and gracefully managing errors.
  • data_visualization: Generates visual representations of data.

---

Installation

1. Clone this repository:

   git clone [https://github.com/MiguelAutomate/AutoGenStudio-Custom-Agents.git](https://github.com/MiguelAutomate/AutoGenStudio-Custom-Agents.git)
   cd AutoGenStudio-Custom-Agents

2. Follow setup instructions: Refer to the SETUP.md file in the cloned repository for detailed instructions on setting up your AutoGen Studio environment and importing these custom components.

---

Usage Examples

Here are a few ways to leverage these custom agents and tools in your AutoGen Studio projects:

Example 1: Using the config_expert to Validate Agent Setup

Let's say you've defined a new agent and want to ensure its configuration is valid.

# In your AutoGen Studio environment or script
from autogen.agentchat.contrib.agent_builder import AgentBuilder

# Assuming you have an agent config dictionary named 'my_new_agent_config'
# e.g., my_new_agent_config = {"name": "TestAgent", "llm_config": {"config_list": [{"model": "gpt-4o-mini"}]}}

# Load the config_expert agent (assuming it's loaded in your AGS instance)
config_expert = autogen.get_agent_by_name("config_expert") # Or initialize directly if not loaded

user_proxy.initiate_chat(
    config_expert,
    message=f"Validate this agent configuration for me: {my_new_agent_config}"
)