OutputConfig

Configure agent output behavior including verbosity, streaming, and formatting

Control how agents display output, from silent mode for programmatic use to verbose mode with rich formatting.

Warning

verbose=, stream=, and metrics= are no longer accepted as top-level Agent kwargs. A top-level verbose=True raises TypeError, and the error names the fix:

text

TypeError: Agent.__init__() got unexpected keyword argument(s): verbose
  verbose: verbosity moved into output=; use output='verbose'
           (or output=OutputConfig(verbose=True)).

Use a preset like output="verbose", or output=OutputConfig(...) for fine-grained control.

python

# Old — raises TypeError on the current SDK
agent = Agent(name="X", verbose=True, stream=True, metrics=True)

# New — preset
agent = Agent(name="X", output="verbose")

# New — dataclass
from praisonaiagents.config import OutputConfig
agent = Agent(name="X", output=OutputConfig(verbose=True, stream=True, metrics=True))

Diagram

Quick Start

Default (Silent Mode)

Silent mode is the default for programmatic use:

python

from praisonaiagents import Agent

# No output overhead - fastest performance
agent = Agent(
    name="Silent Agent",
    instructions="Work quietly"
)

With Presets

Use string presets for common configurations:

python

from praisonaiagents import Agent

# Actions mode - shows tool calls
agent = Agent(
    name="Traced Agent",
    instructions="Show what I do",
    output="actions"
)

# Verbose mode - rich panels
agent = Agent(
    name="Verbose Agent",
    instructions="Show everything",
    output="verbose"
)

With Configuration

Fine-grained control:

python

from praisonaiagents import Agent
from praisonaiagents.config import OutputConfig

agent = Agent(
    name="Custom Output Agent",
    instructions="Custom output settings",
    output=OutputConfig(
        verbose=True,
        markdown=True,
        stream=True,
        metrics=True
    )
)

Configuration Options

python

from praisonaiagents.config import OutputConfig

config = OutputConfig(
    # Verbosity
    verbose=False,

    # Formatting
    markdown=False,

    # Streaming
    stream=False,

    # Metrics display
    metrics=False,

    # Show reasoning steps
    reasoning_steps=False,

    # Output style
    style=None,

    # Actions trace mode
    actions_trace=False,

    # JSON output mode
    json_output=False,

    # Simple output (no panels)
    simple_output=False,

    # Show LLM parameters (debug)
    show_parameters=False,

    # Status trace mode
    status_trace=False,

    # Editor mode (numbered steps)
    editor_output=False,

    # Save response to file
    output_file=None,

    # Response format template
    template=None,

    # Max characters for tool output
    tool_output_limit=16000
)

Parameter	Type	Default	Description
`verbose`	`bool`	`False`	Enable verbose output
`markdown`	`bool`	`False`	Format output as markdown
`stream`	`bool`	`False`	Stream output tokens
`metrics`	`bool`	`False`	Show performance metrics
`reasoning_steps`	`bool`	`False`	Display reasoning process
`style`	`Any \	None`	`None`	Custom output styling
`actions_trace`	`bool`	`False`	Show tool calls and lifecycle
`json_output`	`bool`	`False`	Emit JSONL events
`simple_output`	`bool`	`False`	Plain text without panels
`show_parameters`	`bool`	`False`	Show LLM parameters (debug)
`status_trace`	`bool`	`False`	Inline status updates
`editor_output`	`bool`	`False`	Beginner-friendly numbered steps
`output_file`	`str \	None`	`None`	Save response to file
`template`	`str \	None`	`None`	Response format template
`tool_output_limit`	`int`	`16000`	Max characters per tool output

Output Presets

Eight primary presets cover the common cases. Default is "silent".

Preset	Description	Equivalent `OutputConfig(...)`
`"silent"`	No output (default, fastest)	all flags `False`
`"status"`	Tool calls without timestamps	`actions_trace=True, simple_output=True`
`"trace"`	Full execution trace with timestamps	`actions_trace=True, status_trace=True`
`"verbose"`	Detailed with rich panels	`verbose=True, markdown=True`
`"debug"`	Trace + metrics + reasoning	`metrics=True, reasoning_steps=True, actions_trace=True, status_trace=True, show_parameters=True`
`"stream"`	Real-time token streaming	`verbose=True, markdown=True, stream=True`
`"json"`	Machine-readable JSONL events	`actions_trace=True, json_output=True`
`"editor"`	Beginner-friendly numbered steps	`editor_output=True`

Aliases

Aliases map to a primary preset for backward compatibility.

Alias	Maps To
`"plain"`	`"silent"`
`"minimal"`	`"silent"`
`"normal"`	`"verbose"`
`"actions"`	`"status"`
`"text"`	`"status"`

Common Patterns

Pattern 1: Streaming Chat

python

from praisonaiagents import Agent
from praisonaiagents.config import OutputConfig

agent = Agent(
    name="Chat Agent",
    instructions="Interactive chat",
    output=OutputConfig(
        stream=True,
        markdown=True,
        simple_output=True
    )
)

Pattern 2: Save to File

python

from praisonaiagents import Agent
from praisonaiagents.config import OutputConfig

agent = Agent(
    name="Writer Agent",
    instructions="Generate content",
    output=OutputConfig(
        output_file="output.md",
        template="# {title}\n\n{content}"
    )
)

Pattern 3: JSON Pipeline

python

from praisonaiagents import Agent
from praisonaiagents.config import OutputConfig

agent = Agent(
    name="Pipeline Agent",
    instructions="Emit structured events",
    output=OutputConfig(json_output=True)
)

Best Practices

Use Silent Mode for Production

Silent mode has zero output overhead, making it ideal for programmatic use.

Use Actions Mode for Debugging

Actions trace shows tool calls and agent lifecycle without full verbosity.

Enable Streaming for Interactive Use

Streaming improves perceived responsiveness for chat interfaces.

Display System

Learn about the display system

Output Styles

Customize output styling

OutputConfig

Quick Start

Configuration Options

Output Presets

Aliases

Common Patterns

Pattern 1: Streaming Chat

Pattern 2: Save to File

Pattern 3: JSON Pipeline

Best Practices

Related