Agents Manager

An intelligent multi-agent system that automatically selects the best agent and tool sequence to handle user queries. The system uses semantic similarity to match user requests with specialized agents, each equipped with specific tools for different tasks.

Features

• Intelligent Agent Selection: Uses cosine similarity with embeddings to find the most relevant agent for each query
• Multi-Tool Orchestration: Automatically determines the optimal sequence of tools to complete tasks
• Conversation Context: Maintains conversation history to provide context-aware responses
• Specialized Agents: Five distinct agents, each with specialized capabilities:
  • CSV Data processing and analysis
  • Emoji content creation and translation
  • Text summarization and expansion
  • PDF content extraction and reformatting
  • Multilingual translation with style preservation

Quick Start

Prerequisites

• Python 3.12+
• UV package manager

Installation

1. Install dependencies:

   uv sync

2. Environment setup: Create a .env file with your OpenAI configuration:

    OPENAI_API_KEY=your_api_key_here
    BASE_URL_OPENAI= "https://api.openai.com/v1"
    MODEL_NAME= "gpt-4o"
    EMBEDDING_MODEL_NAME ="text-embedding-3-large"

   See apicall.py for configuration details.

3. Run the application:

   # Using UV
   uv run final_version.py
   
   # Or activate the environment and run directly
   source .venv/bin/activate  # or .venv\Scripts\activate on Windows
   python final_version.py

How It Works

Core Workflow

1. User Query Processing: The system receives and processes user input
2. Agent Selection: Uses semantic similarity to find the most relevant agent
3. Tool Sequence Planning: Determines the optimal sequence of tools to use
4. Tool Execution: Executes tools in sequence, passing context between them
5. Result Generation: Provides comprehensive results to the user

List of Agents and Tools

Available Agents

1. CSVDataAgent

Purpose: CSV data loading, processing, and analysis

• DataframeLoader: Loads CSV/Excel files into pandas DataFrames
• AdvanceCSVQuery: Performs complex queries and filtering on CSV data

2. EmojiWriter

Purpose: Emoji content creation and translation

• EmojiTranslator: Converts text and concepts to relevant emojis
• EmojiMixer: Creates custom emoji combinations from multiple concepts

3. TextSummarizer

Purpose: Text analysis and content expansion

• KeypointExtractor: Identifies key points and main ideas from text
• ContentExpander: Expands bullet points into full paragraphs

4. PDFContentWriter

Purpose: PDF processing and content reformatting

• TextExtractor: Extracts text content from PDF files
• ContentReformatter: Reformats content into various document types

5. TranslatorAgent

Purpose: Multilingual translation with style preservation

• MultilingualTranslator: Translates content between multiple languages
• StylePreserver: Maintains writing style during translation

Architecture Overview

Query Processing Flow

Tool Execution

Flow Diagram

flowchart TD
    subgraph UserInteraction
        A[User] -->|"Enter Query"| B[process_user_query]
    end
    
    subgraph AgentSelection
        B -->|"Store in"| C[ConversationHistory]
        B -->|"Find relevant"| D[find_relevant_agents]
        D -->|"Uses"| E[CosineSimilarityCalculator]
        D -->|"Returns"| F[Top Agent]
    end
    
    subgraph ToolSelection
        F -->|"Input to"| G[determine_tool_sequence]
        G -->|"Tool sequence"| H[Tool Execution Loop]
    end
    
    subgraph ToolExecution
        H -->|"For each tool"| I[gather_tool_inputs]
        I -->|"Inputs for"| J[execute_tool]
        J -->|"Store results"| K[Results Dictionary]
        K -->|"Generate"| L[create_results_summary]
    end

    subgraph DataFlow
        C -.->|"Context for"| I
        C -.->|"Store outputs"| J
        J -.->|"Provide context"| I
    end
    
    subgraph ToolImplementations
        J -->|"Uses"| T1[DataframeLoader]
        J -->|"Uses"| T2[ContentExtractor]
        J -->|"Uses"| T3[EmojiTranslator]
        J -->|"Uses"| T4[EmojiMixer]
        J -->|"Uses"| T5[KeypointExtractor]
        J -->|"Uses"| T6[ContentExpander]
        J -->|"Uses"| T7[TextExtractor]
        J -->|"Uses"| T8[ContentReformatter]
        J -->|"Uses"| T9[MultilingualTranslator]
        J -->|"Uses"| T10[StylePreserver]
    end
    
    subgraph Embedding
        B -.->|"Prepare agents"| Z[generate_agent_embeddings]
        Z -.->|"Creates"| Y[agents_with_embeddings]
        D -.->|"Uses"| Y
    end
    
    L -->|"Display"| A

    style A fill:#f9f,stroke:#333,stroke-width:2px
    style F fill:#bbf,stroke:#333,stroke-width:2px
    style J fill:#bfb,stroke:#333,stroke-width:2px
    style L fill:#ffb,stroke:#333,stroke-width:2px

Loading

Key Components

Core Files

• final_version.py: Main application entry point with agent selection logic
• tools.py: Implementation of all tools used by agents
• agents_behaviour.yaml: Configuration file defining agents and their capabilities
• apicall.py: OpenAI API integration and embedding generation
• prompts.py: Prompt templates for various system interactions

Configuration

The system is configured through agents_behaviour.yaml, which defines:

• Agent descriptions and capabilities
• Tool specifications and versions
• Input/output schemas for each tool
• Agent-specific parameters

Conversation Management

• Maintains conversation history in conversation_history.json
• Provides context to subsequent tool executions
• Enables multi-turn conversations with memory

Example Usage

# Example queries the system can handle:

# CSV Data Analysis
"Load the earthquake.csv file and show me the top 10 earthquakes by magnitude"

# Content Translation
"Translate this text to Spanish: 'Hello, how are you today?'"

# PDF Processing
"Extract text from my resume.pdf and reformat it as a LinkedIn summary"

# Emoji Content
"Convert this sentence to emojis: 'I love sunny days at the beach'"

# Text Summarization
"Extract key points from this article and expand them into a full summary"

Technical Details

Dependencies

• Pandas: Data manipulation and analysis
• OpenAI: Language model integration
• PyMuPDF: PDF text extraction
• Pydantic: Data validation and type safety
• NumPy: Numerical computations
• PyYAML: Configuration file parsing

Similarity Threshold

• Agents are selected only if similarity score > 0.5
• If no agent meets the threshold, falls back to general chat

Extensibility

• Easy to add new agents by updating agents_behaviour.yaml
• New tools can be implemented in tools.py
• Modular architecture supports rapid development

Development

Project Structure

agents_manager/
├── final_version.py          # Main application
├── tools.py                  # Tool implementations
├── agents_behaviour.yaml     # Agent configuration
├── apicall.py               # OpenAI integration
├── prompts.py               # Prompt templates
├── pyproject.toml           # Project dependencies
├── docs/                    # Documentation
├── images/                  # Flow diagrams
└── tool/                    # Additional tool scripts

Adding New Agents

1. Define agent in agents_behaviour.yaml
2. Implement tools in tools.py
3. Update tool mapping in final_version.py
4. Test with relevant queries

Troubleshooting

Common Issues

• "No relevant agent found": Ensure your query matches one of the agent descriptions
• API errors: Check your OpenAI API configuration in .env
• File path issues: Use absolute paths or ensure files are in the working directory

Debug Mode

Set conversation history to see detailed execution flow in conversation_history.json

Contributing

1. Fork the repository
2. Create a feature branch
3. Add your agent/tool implementation
4. Update documentation
5. Submit a pull request

License

This project is available under the MIT License.