Reasoning Display and Logging

MinimalAgent provides visibility into the agent's thinking process through both visual output and programmatic access to reasoning data.

Visual Reasoning Display

The agent can display its reasoning process directly in the terminal:

from minimalagent import Agent, tool

@tool
def search(query: str) -> dict:
    """Search for information."""
    return {"results": f"Results for {query}"}

# By default, colorized reasoning is shown in the terminal
agent = Agent(tools=[search])
response, reasoning = agent.run("Find information about machine learning")

The display includes: - The original query - Each thinking step - Tools being used with their inputs - Tool results - The final response

This makes it easy to understand how the agent reached its conclusion and which tools it used.

Controlling the Display

You can enable or disable the colorized reasoning display:

# Enable reasoning display (default)
agent = Agent(
    tools=[search],
    show_reasoning=True
)

# Disable reasoning display for production
agent = Agent(
    tools=[search],
    show_reasoning=False
)

Accessing Reasoning Data Programmatically

Even when visual display is disabled, you can access the complete reasoning data:

# Disable visual display but still capture reasoning data
agent = Agent(tools=[search], show_reasoning=False)
response, reasoning = agent.run("Search for quantum computing")

# Access reasoning data programmatically
print(f"Query: {reasoning.query}")
print(f"Steps taken: {reasoning.total_steps}")

# Examine each step
for step in reasoning.steps:
    print(f"Step {step.step_number} thinking: {step.thinking}")

    # Examine tools used
    for tool in step.tools:
        print(f"Tool: {tool.name}")
        print(f"Inputs: {tool.inputs}")
        print(f"Result: {tool.result}")

This allows you to build your own visualizations or analytics on top of the reasoning data.

Agent Log Levels

The agent's log level controls what information is logged during operation:

# Minimal logging (warnings and errors only)
agent = Agent(
    tools=[search],
    log_level="WARNING"  # Default
)

# Informational logs (good for general use)
agent = Agent(
    tools=[search],
    log_level="INFO"
)

# Verbose debugging (most detailed)
agent = Agent(
    tools=[search],
    log_level="DEBUG"
)

Available log levels from least to most verbose: - "CRITICAL" - Only critical errors - "ERROR" - Error messages - "WARNING" - Warnings and errors (default) - "INFO" - General information plus warnings and errors - "DEBUG" - Detailed debug information plus everything above

Common Logging Scenarios

Development/Debugging

For maximum visibility during development:

agent = Agent(
    tools=[search],
    show_reasoning=True,  # See colorized reasoning
    log_level="DEBUG"     # Maximum logging detail
)

Production

For production environments:

agent = Agent(
    tools=[search],
    show_reasoning=False,  # No visual display
    log_level="WARNING"    # Only important warnings and errors
)

Monitoring

For monitoring in production while maintaining some visibility:

agent = Agent(
    tools=[search],
    show_reasoning=False,  # No visual display
    log_level="INFO"       # Information about operations, warnings, and errors
)

Advanced: Integration with Python Logging

MinimalAgent integrates with Python's standard logging system, allowing you to capture logs in your application's logging infrastructure:

import logging

# Configure Python's logging system
logging.basicConfig(
    level=logging.INFO,
    format="%(asctime)s - %(name)s - %(levelname)s - %(message)s",
    filename="agent.log"  # Save to a file instead of console
)

# MinimalAgent will use this logging configuration
agent = Agent(
    tools=[search],
    log_level="INFO",  # This level applies to MinimalAgent's logger
)

This is useful for: - Saving logs to files instead of the console - Integrating with existing logging infrastructure - Custom log formatting - Log rotation and management