SideSeat Class

The SideSeat class is the main entry point for the Python SDK. It manages OpenTelemetry setup, framework instrumentation, and exporters.

Class Definition

class SideSeat:
    def __init__(
        self,
        endpoint: str | None = None,
        project_id: str | None = None,
        api_key: str | None = None,
        framework: str | None = None,
        auto_instrument: bool = True,
        service_name: str | None = None,
        service_version: str | None = None,
        enable_traces: bool = True,
        enable_metrics: bool = True,
        enable_logs: bool = False,
        capture_content: bool = True,
        encode_binary: bool = True,
        disabled: bool = False,
        debug: bool = False,
    ) -> None: ...

Properties

config

Immutable configuration object.

client = SideSeat()
print(client.config.endpoint)
print(client.config.project_id)

telemetry

Access to the underlying telemetry client for debug exporters.

client = SideSeat()
client.telemetry.setup_console_exporter()
client.telemetry.setup_file_exporter("traces.jsonl")

tracer_provider

The OpenTelemetry TracerProvider instance managing trace collection.

client = SideSeat()
provider = client.tracer_provider

# Force flush all pending spans
provider.force_flush()

is_disabled

Whether telemetry is disabled.

client = SideSeat()
if client.is_disabled:
    print("Telemetry is disabled")

Methods

span(name)

Create a custom span for tracing.

Parameters:

• name (str): Span name

Returns: ContextManager[Span]

client = SideSeat()

with client.span("my-operation") as span:
    span.set_attribute("user_id", "123")
    result = process_request()
# Errors are automatically recorded with stack trace

get_tracer(name)

Get an OpenTelemetry tracer.

Parameters:

• name (str): Tracer name

Returns: Tracer

client = SideSeat()
tracer = client.get_tracer("my-module")

with tracer.start_as_current_span("operation"):
    # ... work ...
    pass

force_flush(timeout_millis)

Export pending spans immediately.

Parameters:

• timeout_millis (int): Timeout in milliseconds. Default: 30000

Returns: bool

client = SideSeat()
# ... do work ...
success = client.force_flush(timeout_millis=5000)

validate_connection(timeout)

Test server connectivity.

Parameters:

• timeout (float): Timeout in seconds. Default: 5.0

Returns: bool

client = SideSeat()
if client.validate_connection():
    print("Server is reachable")

shutdown(timeout_millis)

Gracefully shutdown all exporters, flushing pending spans.

Parameters:

• timeout_millis (int): Timeout in milliseconds. Default: 30000

Returns: None

client = SideSeat()
# ... do work ...
client.shutdown()

Tip

shutdown() is:

• Idempotent: Safe to call multiple times
• Thread-safe: Can be called from any thread
• Automatic: Registered with atexit, so you don’t have to call it

Context Manager

Use the context manager for automatic shutdown:

with SideSeat() as client:
    # Your code runs with tracing enabled
    run_my_agent()
# Traces flushed and shutdown automatically

Usage Patterns

Basic Usage

from sideseat import SideSeat

client = SideSeat()

# Your code runs with tracing enabled
# ...

# Optional: explicit shutdown
client.shutdown()

With Framework Selection

from sideseat import SideSeat, Frameworks

client = SideSeat(framework=Frameworks.Strands)

With Custom Configuration

client = SideSeat(
    endpoint="http://localhost:5388",
    project_id="my-project",
    service_name="my-agent",
    service_version="1.0.0",
)

With Debug Exporters

client = SideSeat()
client.telemetry.setup_console_exporter()  # Print spans
client.telemetry.setup_file_exporter("traces.jsonl")  # Write to file

Disabled Mode

# For testing/CI
client = SideSeat(disabled=True)
# Or: SIDESEAT_DISABLED=true

Thread Safety

The SideSeat class is thread-safe:

• Multiple threads can create spans concurrently
• shutdown() can be called from any thread
• Exporters use internal locking

import threading

client = SideSeat()

def worker():
    with client.span("worker"):
        # ... work ...
        pass

threads = [threading.Thread(target=worker) for _ in range(10)]
for t in threads:
    t.start()
for t in threads:
    t.join()

client.shutdown()

Next Steps

• Configuration — all configuration options
• Exporters — debug exporters for development