Real-time agent that executes tools during streaming responses. Demonstrates building an interactive agent with the Inference Gateway SDK.

# Set environment (optional)
export LLM_PROVIDER="deepseek"
export LLM_MODEL="deepseek-chat"

# Run the agent
go run agent.go

• Real-time execution: Tools execute immediately when JSON is complete
• Multi-tool support: Weather lookup + calculator
• Iterative conversations: Queue-based agent pattern
• Streaming reasoning: See the AI's thinking process

• get_current_weather: Get weather for major cities
• calculate: Basic arithmetic (add, subtract, multiply, divide)

👤 User: What's the weather in San Francisco and what's 15 multiplied by 7?

🤖 Assistant:
🔧 Executing: get_current_weather({"location": "san francisco"})
📋 Result: {"temperature":14,"unit":"celsius","description":"Foggy"}
🔧 Executing: calculate({"a": 15, "b": 7, "operation": "multiply"})
📋 Result: {"operation":"15.00 multiply 7.00","result":105}

The weather in San Francisco is foggy at 14°C.
The result of 15 × 7 is 105.

1. Process user message queue
2. Stream LLM response with tools
3. Execute tools when JSON is complete
4. Add results to conversation history
5. Continue until no more tool calls