damasqas-livekit 0.3.1

Voice AI infrastructure monitoring for LiveKit agents

damasqas_livekit

Voice AI infrastructure monitoring for LiveKit agents. One line to monitor everything.

Installation

pip install damasqas_livekit

Usage

import damasqas_livekit

# After starting your AgentSession:
damasqas_livekit.monitor(session)

Set DAMASQAS_TOKEN as an environment variable (your workspace API key).

What It Captures

• Pipeline metrics — STT, LLM, TTS latencies per turn (grouped by speech_id)
• Tool calls — function name, duration, success/failure
• Transcripts — user and agent speech text
• State transitions — agent lifecycle (listening, thinking, speaking)
• Session close — end reason, errors, total turns

All telemetry is batched and sent asynchronously. The SDK never blocks or crashes your voice pipeline.

Configuration

Environment Variable	Required	Default
DAMASQAS_TOKEN	Yes	—
DAMASQAS_URL	No	https://connect.damasqas.com/api/v1/voice/ingest

Or pass them directly:

damasqas_livekit.monitor(session, token="your-token", url="https://...")

Logging

The SDK is quiet by default — only warnings and errors are logged. To see per-event INFO logs (batch flushes, metric captures, transcript arrivals, etc.) for debugging, pass verbose=True:

damasqas_livekit.monitor(session, verbose=True)