Does this skill support cost tracking for specific models?

Yes, it includes patterns for estimating USD costs based on token usage for models like Claude 3.5 Sonnet and allows for easy updates to pricing logic.

Is this compatible with web frameworks like FastAPI?

Absolutely. It includes specific middleware implementations for FastAPI to handle request correlation and automatic instrumentation.

Does it support distributed tracing?

Yes, it leverages OpenTelemetry to provide full span management, allowing you to trace requests across microservices and external AI providers.

How does it handle sensitive prompt data?

The skill promotes security best practices by including placeholders and patterns for PII redaction to ensure sensitive data isn't leaked into your logs.

Observability and Logging for AI

Name: Observability and Logging for AI
Author: ricardoroche

byricardoroche

分析与监控

Implements production-grade structured logging, OpenTelemetry tracing, and LLM-specific metrics tracking for AI applications.

关于

This skill provides a comprehensive framework for instrumenting AI and LLM applications with robust observability patterns. It automates the implementation of structured JSON logging, distributed tracing via OpenTelemetry, and granular metrics collection for tokens, cost, and latency. By using this skill, developers can ensure their AI workflows are measurable and debuggable, utilizing request correlation to track complex interactions across distributed systems and implementing PII redaction to maintain data privacy in logs.

主要功能

Prometheus metrics collection for real-time operational monitoring
Structured JSON logging for machine-readable diagnostic data
LLM-specific metrics including token usage, USD cost estimation, and latency
OpenTelemetry integration for distributed tracing and span management
0 GitHub stars
Request correlation across asynchronous contexts and API middleware

使用场景

Monitoring and optimizing LLM operational costs and token consumption
Debugging complex multi-step AI agent workflows with request IDs
Setting up production-ready observability dashboards for AI-powered services

关于

主要功能

Prometheus metrics collection for real-time operational monitoring
Structured JSON logging for machine-readable diagnostic data
LLM-specific metrics including token usage, USD cost estimation, and latency
OpenTelemetry integration for distributed tracing and span management
0 GitHub stars
Request correlation across asynchronous contexts and API middleware

使用场景

Monitoring and optimizing LLM operational costs and token consumption
Debugging complex multi-step AI agent workflows with request IDs
Setting up production-ready observability dashboards for AI-powered services