Message Formatting

messages = [
    {"role": "user", "content": "What's the weather in Tokyo?"},
    {"role": "assistant", "content": "Let me check that for you..."},
    {"role": "user", "content": "Tool result: 22°C, sunny"},
    {"role": "assistant", "content": "It's currently 22°C and sunny in Tokyo!"}
]

I need you to:
1. Search for all products in the "Electronics" category
2. Filter to items with less than 10 units in stock
3. For each item, check if there are pending orders
4. Create a restock report with product name, current stock, and pending orders

We just launched a new pricing structure last week. When looking up 
product prices, use the current catalog—don't rely on any cached data. 
What's the current price for product SKU-789?

Please analyze this data:

| Product | Q1 Sales | Q2 Sales |
|---------|----------|----------|
| Widget  | 1,234    | 1,567    |
| Gadget  | 892      | 1,103    |

Focus on:
- Quarter-over-quarter growth percentage
- Which product performed better

messages.append({
    "role": "user", 
    "content": "Tool result for get_weather: Temperature: 22°C, Conditions: Sunny, Humidity: 45%"
})

tool_result = {
    "tool": "search_products",
    "results": [
        {"id": "123", "name": "Wireless Mouse", "price": 29.99, "stock": 45},
        {"id": "456", "name": "USB Keyboard", "price": 49.99, "stock": 12}
    ],
    "total_matches": 2
}

messages.append({
    "role": "user",
    "content": f"Tool result: {json.dumps(tool_result, indent=2)}"
})

messages.append({
    "role": "user",
    "content": """Tool 'get_order' failed:
Error: Order not found
Details: No order exists with ID ORD-99999
Suggestion: Verify the order ID or search by customer email instead"""
})

# Initial request
messages = [{"role": "user", "content": "Book a meeting with the team"}]

# Agent asks for clarification
response = client.agent(messages=messages, ...)
# Response: "I'd be happy to book a meeting. Could you tell me:
# 1. Which team members should I invite?
# 2. What date and time works best?
# 3. How long should the meeting be?"

# User provides details
messages.append({"role": "assistant", "content": response.response})
messages.append({"role": "user", "content": """
- Invite: [email protected], [email protected], [email protected]
- Date: Next Tuesday at 2pm
- Duration: 1 hour
"""})

messages = [
    {"role": "user", "content": "Search for laptops under $1000"},
    {"role": "assistant", "content": "I found 15 laptops. Here are the top 5..."},
    {"role": "user", "content": "Show me only the ones with at least 16GB RAM"},
    {"role": "assistant", "content": "Narrowing to 16GB+ RAM, here are 6 options..."},
    {"role": "user", "content": "Compare the top 2 in a table format"}
]

# Step 1: User request
messages = [{"role": "user", "content": "Get the weather in all cities where we have offices"}]

# Step 2: Agent calls get_offices tool first
# After getting offices: ["NYC", "London", "Tokyo"]

# Step 3: Provide office results
messages.append({"role": "assistant", "content": "Let me first find our office locations..."})
messages.append({"role": "user", "content": "Tool result for get_offices: NYC, London, Tokyo"})

# Step 4: Agent now calls get_weather for each city
# Continue the loop...

Extract structured data from these customer reviews.

Format each review as:
{
  "sentiment": "positive" | "negative" | "neutral",
  "main_topic": "string",
  "key_phrases": ["string", ...]
}

Example input: "Great product, fast shipping!"
Example output: {"sentiment": "positive", "main_topic": "satisfaction", "key_phrases": ["great product", "fast shipping"]}

Now process these reviews:
1. "The quality is poor and it arrived late."
2. "Works as expected, nothing special."
3. "Best purchase I've made this year!"

# Instead of keeping 50 messages, summarize
messages = [
    {
        "role": "user", 
        "content": """Previous conversation summary:
- User asked about Q3 sales data
- Found total revenue of $1.2M
- Identified top 3 products by volume
- User requested comparison with Q2

Current request: Show me the growth percentage for each product category."""
    }
]

# ❌ Including unnecessary verbosity
{"role": "user", "content": f"Tool result: {huge_json_blob_with_all_fields}"}

# ✅ Include only relevant data
{"role": "user", "content": f"Tool result: Found 3 matching orders. Total value: $1,234.56"}