from incredible import AgentMax

# Initialize
agent = AgentMax(api_key="YOUR_API_KEY")

# Use run_with_results() to get structured output back
result = agent.run_with_results(
    goal="Research competitors and summarize findings",
    result_structure={
        "type": "object",
        "properties": {
            "competitors": {"type": "array"},
            "summary": {"type": "string"}
        }
    }
)

print(result.output)  # Guaranteed to match your schema

# run() is for fire-and-forget tasks with side effects
agent.run(
    goal="Send personalized outreach emails to all at-risk customers",
    files=[open("customers.csv", "rb")],  # Pass file objects
    data={"context": "Q1 2024"},
    tools=[crm_tool, email_tool],
    max_steps=50,
    timeout=300
)

# The task runs. Emails get sent. No structured output returned.
print("Outreach campaign started!")

# Open files properly
with open("meeting_transcript.txt", "r") as f:
    result = agent.run_with_results(
        goal="Extract all action items from this meeting",
        files=[f],
        result_structure={
            "type": "object",
            "properties": {
                "action_items": {
                    "type": "array",
                    "items": {
                        "type": "object",
                        "properties": {
                            "task": {"type": "string"},
                            "assignee": {"type": "string"},
                            "due_date": {"type": "string"},
                            "priority": {"type": "string", "enum": ["high", "medium", "low"]}
                        },
                        "required": ["task", "assignee"]
                    }
                },
                "summary": {"type": "string"}
            },
            "required": ["action_items", "summary"]
        }
    )

# result.output ALWAYS matches the schema
for item in result.output["action_items"]:
    print(f"- [{item['priority']}] {item['task']} → {item['assignee']}")

for event in agent.stream(
    goal="Write a detailed market analysis"
):
    if event.type == "step":
        print(f"Step {event.step}: {event.action}")
    elif event.type == "progress":
        print(f"Progress: {event.message}")
    elif event.type == "output":
        print(f"Output chunk: {event.content}")
    elif event.type == "done":
        print(f"Complete! Total steps: {event.total_steps}")

from incredible import Tool

# Define a tool
crm_search = Tool(
    name="search_crm",
    description="Search the CRM for customer information",
    parameters={
        "type": "object",
        "properties": {
            "query": {
                "type": "string",
                "description": "Search query (company name, email, etc.)"
            },
            "limit": {
                "type": "integer",
                "description": "Max results to return",
                "default": 10
            }
        },
        "required": ["query"]
    },
    # Handler function that executes when tool is called
    handler=lambda params: your_crm_api.search(params["query"], params.get("limit", 10))
)

# Use the tool with run_with_results to get data back
result = agent.run_with_results(
    goal="Find all customers in the healthcare industry",
    tools=[crm_search],
    result_structure={
        "type": "object",
        "properties": {
            "customers": {"type": "array"},
            "total_count": {"type": "integer"}
        }
    }
)

# Open files as proper file objects
with open("report.pdf", "rb") as pdf, \
     open("data.csv", "r") as csv, \
     open("screenshot.png", "rb") as img:
    
    result = agent.run_with_results(
        goal="Summarize these documents",
        files=[pdf, csv, img],
        result_structure={
            "type": "object",
            "properties": {
                "summary": {"type": "string"}
            }
        }
    )

# Using pathlib
from pathlib import Path

pdf_file = Path("report.pdf").open("rb")
result = agent.run_with_results(
    goal="Extract key findings",
    files=[pdf_file],
    result_structure={"type": "object", "properties": {"findings": {"type": "array"}}}
)
pdf_file.close()

result = agent.run_with_results(
    goal="Research latest AI news and summarize top 5 stories",
    builtin_tools=["web_search", "web_scrape"],
    result_structure={
        "type": "object",
        "properties": {
            "stories": {"type": "array"},
            "summary": {"type": "string"}
        }
    }
)

from incredible import AgentMax, AgentMaxError, TimeoutError, ValidationError

agent = AgentMax(api_key="YOUR_API_KEY")

try:
    result = agent.run(goal="Complex task")
except TimeoutError:
    print("Task took too long")
except ValidationError as e:
    print(f"Invalid input: {e.message}")
except AgentMaxError as e:
    print(f"Agent error: {e.message}")
    print(f"Partial results: {e.partial_output}")
    print(f"Steps completed: {e.steps_completed}")

result = agent.run(
    goal="Long research task",
    webhook_url="https://your-app.com/webhook",
    webhook_events=["complete", "error", "step"]
)

# Returns immediately with task ID
print(f"Task started: {result.task_id}")