Strands Agents SDK with Open Web UI - Specialised Agents Edition

This is an advanced version of the Strands Agents integration with Open Web UI, featuring specialised agents for different tasks and tools. Each agent is designed to excel at specific capabilities using dedicated tools from the Strands Agents Tools library.

🎯 What's New in Version 2.0

• 6 Specialised Agents: Each agent focuses on specific tools and tasks
• Organised by Category: Agents are grouped into logical categories (utility, files, web)
• Educational Design: Simple, clear code structure perfect for learning
• OpenAI Compatible: Works seamlessly with Open Web UI and other OpenAI-compatible clients
• Agent Discovery: REST API endpoints to explore available agents and their capabilities
• Extensible Architecture: Ready for additional agent categories

🤖 Currently Implemented Agents

Utility Agents (3 agents)

• Calculator Agent (calculator) - Mathematical computations and calculations
• Python REPL Agent (python-repl) - Python code execution and data analysis
• HTTP Request Agent (http-request) - Web API interactions and HTTP operations

File Management Agents (2 agents)

• File Read Agent (file-read) - Reading and analysing file contents
• File Write Agent (file-write) - Creating and writing file contents

Web & Search Agents (1 agent)

• Tavily Search Agent (tavily-search) - Real-time web search and information retrieval

Plus Alfred Butler

• Alfred Butler (alfred-butler) - Default conversational agent with British personality

Total: 6 specialised agents + Alfred Butler = 7 available models

🔮 Future Agent Categories (Architecture Ready)

The system is designed to easily add agents in these categories:

• AWS Service Agents - Bedrock Memory, AWS operations
• Communication Agents - Slack, RSS, notifications
• Media & Content Agents - Image/video generation, diagrams
• System Agents - Environment management, MCP integration
• Orchestration Agents - Batch processing, workflows

🚀 Quick Start

1. Installation

# Clone the repository
git clone <repository-url>
cd strands-agents-sdk-with-open-web-ui/repo2

# Install dependencies
pip install -r requirements.txt

2. Configuration

Set up your credentials for the required services:

AWS Bedrock (Required)

# Option 1: Bedrock API Key (for development)
export AWS_BEDROCK_API_KEY=your_bedrock_api_key

# Option 2: AWS Credentials (for production)
export AWS_ACCESS_KEY_ID=your_access_key
export AWS_SECRET_ACCESS_KEY=your_secret_key
export AWS_REGION=us-west-2

Tavily API (Required for Tavily Search Agent)

# Get your API key from https://tavily.com/
export TAVILY_API_KEY=your_tavily_api_key

Note: The Tavily Search Agent requires a valid Tavily API key. You can get one for free at https://tavily.com/. Without this key, the Tavily Search Agent will not be available, but all other agents will work normally.

3. Run the API Server

# Start the server
python -m uvicorn src.api:app --host 0.0.0.0 --port 8000

# Or use the development server with auto-reload
python -m uvicorn src.api:app --host 0.0.0.0 --port 8000 --reload

4. Test with Open Web UI

1. Open your Open Web UI instance
2. Add a new connection: http://localhost:8000
3. Select from available models/agents:
   • alfred-butler (default conversational agent)
   • calculator (mathematical computations)
   • python-repl (Python code execution)
   • http-request (web API interactions)
   • And many more!

📚 API Endpoints

OpenAI-Compatible Endpoints

• GET /v1/models - List all available agents as models
• POST /v1/chat/completions - Chat with any agent
• GET /v1/models/{model_id} - Get specific agent/model info

Agent Discovery Endpoints

• GET /v1/agents - List all agents with categories
• GET /v1/agents/{agent_id} - Get detailed agent information
• GET /v1/agents/category/{category} - Get agents by category

Example: Using a Specific Agent

# Chat with the Calculator Agent
curl -X POST "http://localhost:8000/v1/chat/completions" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "calculator",
    "messages": [
      {"role": "user", "content": "What is the square root of 144?"}
    ]
  }'

# Chat with the Python REPL Agent
curl -X POST "http://localhost:8000/v1/chat/completions" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "python-repl",
    "messages": [
      {"role": "user", "content": "Create a simple plot of y = x^2 for x from -5 to 5"}
    ]
  }'

🏗️ Architecture

src/
├── agents/                    # Specialised agent implementations
│   ├── base_agent.py         # Abstract base class for all agents
│   ├── utility/              # General-purpose agents
│   ├── files/                # File management agents
│   ├── web/                  # Web and search agents
│   ├── aws/                  # AWS service agents
│   ├── communication/        # Messaging agents
│   ├── media/                # Content creation agents
│   ├── system/               # System operation agents
│   ├── orchestration/        # Workflow coordination agents
│   └── memory/               # Memory management agents
├── registry/                 # Agent discovery and management
│   └── agent_registry.py     # Central agent registry
├── api.py                    # FastAPI REST endpoints
├── service.py                # Business logic
├── models.py                 # Pydantic data models
└── agent.py                  # Original Alfred agent (still available)

🎓 Educational Features

This repository is designed for learning and understanding:

• Simple Code Structure: Each agent is self-contained and easy to understand
• Clear Documentation: Extensive comments and docstrings in British English
• Modular Design: Easy to add new agents or modify existing ones
• Best Practices: Follows Python and open-source conventions
• Tool Integration: Demonstrates how to use Strands Agents Tools effectively

🔧 Adding New Agents

To add a new specialised agent:

1. Create the agent class in the appropriate category directory:

# src/agents/utility/my_new_agent.py
from strands_tools import my_tool
from ..base_agent import BaseSpecialisedAgent

class MyNewAgent(BaseSpecialisedAgent):
    def get_tools(self):
        return [my_tool]
    
    def get_system_prompt(self):
        return "You are a specialist in..."
    
    def get_agent_name(self):
        return "My New Agent"
    
    def get_agent_description(self):
        return "Specialised agent for..."

2. Register the agent in src/registry/agent_registry.py:

from ..agents.utility.my_new_agent import MyNewAgent

class AgentRegistry:
    _agents = {
        # ... existing agents
        "my-new-agent": MyNewAgent,
    }

3. Test your agent:

# List all agents to verify registration
curl http://localhost:8000/v1/agents

# Test your new agent
curl -X POST "http://localhost:8000/v1/chat/completions" \
  -H "Content-Type: application/json" \
  -d '{"model": "my-new-agent", "messages": [{"role": "user", "content": "Hello!"}]}'

🆚 Comparison with Simple Version

Feature	Simple Version (repo/)	Specialised Version (repo2/)
Agents	1 (Alfred Butler)	6 Specialised + Alfred
Tools	None	Calculator, Python, HTTP, Files, Search
Organisation	Single file	Categorised by function
Discovery	Fixed model	Dynamic agent registry
Use Case	General conversation	Task-specific expertise
Learning	Basic integration	Advanced architecture

🛠️ Development

Running Tests

# Install development dependencies
pip install -r requirements.dev.txt

# Run tests
python -m pytest tests/

# Run with coverage
python -m pytest tests/ --cov=src

Code Quality

# Format code
black src/ tests/

# Lint code
flake8 src/ tests/

# Type checking
mypy src/

📄 License

This project is licensed under the MIT License - see the LICENSE.md file for details.

🤝 Contributing

Contributions are welcome! Please read our contributing guidelines and submit pull requests for any improvements.

🔗 Related Projects

• Strands Agents SDK - The core SDK
• Strands Agents Tools - Community tools library
• Open Web UI - The web interface

---

Note: This is the advanced version with specialised agents. For a simpler implementation, see the repository which contains the basic Alfred Butler agent.