Does it support monitoring for voice-specific features?

Absolutely. It includes specific span kinds for STT transcription, TTS synthesis, VAD audio levels, and full voice session lifecycles.

Can I filter out high-frequency telemetry events?

Yes, the TelemetryConfig allows you to suppress specific hook triggers, such as audio level updates, to reduce noise in your logs.

Can I track LLM token usage and latency?

Yes, the skill automatically captures span attributes for input/output tokens, tool counts, and time-to-first-byte (TTFB) for LLM generations.

Which telemetry providers are supported by this skill?

It supports three main providers: OpenTelemetry for full SDK integration, ConsoleTelemetryProvider for stdout logging, and NoopTelemetryProvider for zero overhead.

How do I integrate this with Jaeger?

You can configure the OpenTelemetryProvider with a standard OTel TracerProvider and a Jaeger exporter to visualize your RoomKit spans in Jaeger.

RoomKit Telemetry Setup

Name: RoomKit Telemetry Setup
Author: roomkit-live

byroomkit-live

•

分析与监控

Integrates comprehensive observability and OpenTelemetry tracing into RoomKit AI conversation agents to monitor performance and latency.

This skill enables developers to implement robust observability for RoomKit applications, providing deep insights into LLM latency, voice pipeline performance, and message delivery status. It supports multiple telemetry providers, including OpenTelemetry and console logging, allowing users to trace synchronous and asynchronous hooks, monitor speech-to-text (STT) and text-to-speech (TTS) sessions, and capture custom application spans. This is essential for debugging multi-channel AI agents and optimizing the user experience in production environments.

主要功能

01Detailed monitoring of LLM token usage, TTFB, and generation latency

022 GitHub stars

03Fine-grained control over sampling rates and metadata suppression

04Tracing for voice pipelines, including STT transcription and TTS synthesis

05Support for OpenTelemetry, Console, and Noop telemetry providers

06Automated span generation for hook execution and message delivery tracking

使用场景

01Implementing distributed tracing with Jaeger for multi-channel agent architectures

02Debugging latency bottlenecks in real-time AI voice conversation pipelines

03Monitoring LLM consumption and performance metrics across production rooms

主要功能

01Detailed monitoring of LLM token usage, TTFB, and generation latency

022 GitHub stars

03Fine-grained control over sampling rates and metadata suppression

04Tracing for voice pipelines, including STT transcription and TTS synthesis

05Support for OpenTelemetry, Console, and Noop telemetry providers

06Automated span generation for hook execution and message delivery tracking

使用场景

01Implementing distributed tracing with Jaeger for multi-channel agent architectures

02Debugging latency bottlenecks in real-time AI voice conversation pipelines

03Monitoring LLM consumption and performance metrics across production rooms