Silent Mode (no_output)¶

Run Flock quietly when embedded in other applications or deployed as a service.

When using Flock as part of a larger application, you may want to suppress the decorative terminal output (banners, Rich tables, streaming displays) while keeping logs available for debugging.

Quick Start¶

from flock import Flock

# Suppress all decorative terminal output
flock = Flock("openai/gpt-4.1", no_output=True)

That's it! The banner, Rich tables, and streaming displays are now suppressed. Only programmatic output (your own print() statements) and configured logging will appear.

What Gets Suppressed¶

Output Type	Default	With `no_output=True`
Startup banner	✅ Shown	❌ Hidden
Rich streaming display	✅ Shown	❌ Hidden
Agent output tables	✅ Shown	❌ Hidden
Your `print()` statements	✅ Shown	✅ Shown
Logging output	✅ Shown	✅ Shown

Use Cases¶

Running as a Service¶

When Flock is embedded in a web service, the decorative output clutters logs:

from fastapi import FastAPI
from flock import Flock, flock_type
from pydantic import BaseModel

app = FastAPI()

# Silent mode for service deployment
flock = Flock("openai/gpt-4.1", no_output=True)

@flock_type
class Query(BaseModel):
    text: str

@flock_type  
class Response(BaseModel):
    answer: str

agent = flock.agent("responder").consumes(Query).publishes(Response)

@app.post("/ask")
async def ask(query: Query):
    await flock.publish(query)
    await flock.run_until_idle()
    responses = await flock.get_artifacts(Response)
    return {"answer": responses[0].answer}

Batch Processing¶

When running many jobs, the output would be overwhelming:

from flock import Flock

flock = Flock("openai/gpt-4.1", no_output=True)

# Process 1000 items without terminal spam
for item in items:
    await flock.publish(item)

await flock.run_until_idle()

# Your summary output only
print(f"Processed {len(items)} items")

Testing¶

Keep test output clean:

import pytest
from flock import Flock

@pytest.fixture
def silent_flock():
    return Flock("openai/gpt-4.1", no_output=True)

def test_agent_workflow(silent_flock):
    # Tests run without decorative output
    ...

CI/CD Pipelines¶

Avoid cluttering CI logs:

import os
from flock import Flock

# Auto-enable silent mode in CI
no_output = os.getenv("CI", "false").lower() == "true"
flock = Flock("openai/gpt-4.1", no_output=no_output)

Propagation to Components¶

The no_output flag automatically propagates to all engines and components:

from flock import Flock, DSPyEngine

flock = Flock("openai/gpt-4.1", no_output=True)

# Custom engines automatically inherit no_output
custom_engine = DSPyEngine(model="openai/gpt-4.1")

agent = (
    flock.agent("processor")
    .consumes(Input)
    .publishes(Output)
    .with_engines(custom_engine)  # no_output=True is propagated
)

This works for:

✅ Default DSPyEngine
✅ Custom engines added via .with_engines()
✅ Default OutputUtilityComponent
✅ Custom utilities added via .with_utilities()

Combining with Logging¶

Silent mode suppresses decorative output but not logging. Configure logging for production visibility:

import logging
from flock import Flock
from flock.logging.logging import configure_logging

# Enable structured logging
configure_logging(level=logging.INFO)

# Suppress decorative output
flock = Flock("openai/gpt-4.1", no_output=True)

# Logs still appear
# INFO:flock.core.orchestrator:Agent 'processor' executed successfully

Recommended Production Configuration¶

import logging
import os
from flock import Flock
from flock.logging.logging import configure_logging

# Production settings
configure_logging(
    level=logging.INFO if os.getenv("DEBUG") else logging.WARNING
)

flock = Flock(
    os.getenv("DEFAULT_MODEL", "openai/gpt-4.1"),
    no_output=True  # Always silent in production
)

Dashboard Compatibility¶

Silent mode works with the dashboard. The dashboard receives data via WebSocket, not terminal output:

flock = Flock("openai/gpt-4.1", no_output=True)

# Dashboard still works!
await flock.serve(dashboard=True)  # Terminal is quiet, dashboard shows everything

Environment Variable Alternative¶

You can also control output via environment variable in your code:

import os
from flock import Flock

flock = Flock(
    "openai/gpt-4.1",
    no_output=os.getenv("FLOCK_NO_OUTPUT", "false").lower() == "true"
)

Then set in your deployment:

export FLOCK_NO_OUTPUT=true
python my_service.py