Skip to main content
Output controls what agents print, how much detail they show, and whether they stream responses in real time. 
from praisonaiagents import Agent

agent = Agent(
    name="Assistant",
    instructions="You are a helpful assistant.",
    output="verbose",
)

agent.start("Explain how neural networks learn.")
The user runs the agent; output mode controls how much detail prints or streams during the run.

​
Quick Start

1

Level 1 — String (preset)

Pick a named output mode — the shortest way to control what the agent prints.
from praisonaiagents import Agent

agent = Agent(
    name="Assistant",
    instructions="You are a helpful assistant.",
    output="verbose",
)
agent.start("What is the fastest sorting algorithm?")
2

Level 2 — Config class (full control)

Use OutputConfig to combine formatting flags like streaming and markdown.
from praisonaiagents import Agent, OutputConfig

agent = Agent(
    name="Assistant",
    instructions="You are a helpful assistant.",
    output=OutputConfig(
        verbose=True,
        markdown=True,
        stream=True,
    ),
)
agent.start("Write a comparison of React vs Vue.js.")

​
Output Presets

The default is output="silent" — agents return responses without printing anything. This is the fastest mode for programmatic use.

​
How It Works

PhaseWhat happens
1. GenerateAgent processes the request normally
2. RouteResponse is passed to the configured output mode
3. DisplayOutput mode formats and renders (or stays silent)

​
Configuration Options

Full list of options, types, and defaults — OutputConfig
The most common options at a glance:
OptionTypeDefaultDescription
verboseboolFalseShow Rich panels and detailed output
markdownboolFalseRender markdown formatting
streamboolFalseStream tokens as they generate
json_outputboolFalseEmit JSONL events for piping
output_filestr | NoneNoneAuto-save response to file path

​
Common Patterns

​
Pattern 1 — Streaming for real-time UX

from praisonaiagents import Agent, OutputConfig

agent = Agent(
    instructions="You are a writing assistant.",
    output=OutputConfig(stream=True, markdown=True),
)
response = agent.start("Write a short story about a robot learning to paint.")
print(response)

​
Pattern 2 — Save output to file

from praisonaiagents import Agent, OutputConfig

agent = Agent(
    instructions="You are a report writer.",
    output=OutputConfig(output_file="report.md", markdown=True),
)
agent.start("Generate a quarterly sales analysis report.")

​
Pattern 3 — JSON output for pipelines

from praisonaiagents import Agent, OutputConfig

agent = Agent(
    instructions="You are a data extraction agent.",
    output=OutputConfig(json_output=True),
)
agent.start("Extract all company names from this text: Apple launched in 1976, Google in 1998.")

​
Best Practices

output="silent" (the default) adds zero overhead and returns the response as a Python string. Use it for APIs, background jobs, and any code that processes the response programmatically.
Set output="verbose" during development to see tool calls, agent reasoning, and formatted output. Switch back to silent before deploying.
Enable OutputConfig(stream=True) for chat interfaces and terminal apps where users expect to see the response appear word by word, not all at once.
Use output_file="path/to/file.md" to automatically save every response — useful for report generation, batch processing, and audit trails.

Caching — avoid redundant LLM calls

Execution — control iteration limits and rate limiting