What data should I avoid instrumenting?

To minimize noise and protect privacy, the skill recommends avoiding full prompt/response content containing PII, every minor intermediate thought, and user secrets.

Does this support multi-agent systems?

Yes, Tier 3 specifically covers parent-child span relationships, agent handoffs, and tracking delegation reasoning to visualize complex multi-agent interactions.

How does this skill help with AI cost management?

The skill provides specific attribute naming conventions for tracking input/output tokens and calculating USD costs per call, enabling precise financial attribution.

What is the 5-tier implementation framework?

It is a staged approach to observability that starts with foundational basics (SDK init, root spans) and progresses to core tracing, cost attribution, multi-agent coordination, and finally quality evaluation.

Instrumentation Planning

Name: Instrumentation Planning
Author: nexus-labs-automation

bynexus-labs-automation

0•

분석 및 모니터링

Architects comprehensive observability strategies for AI agents using a structured, outcome-focused 5-tier framework.

The Instrumentation Planning skill provides a strategic roadmap for implementing observability in AI agent systems. It guides developers through a five-tier implementation process—moving from foundational error capturing and root spans to advanced multi-agent coordination and automated quality evaluations. By establishing standardized naming conventions for spans and attributes, this skill ensures that agent telemetry provides actionable insights into success rates, latency, token costs, and reasoning logic, helping teams transition from prototype to production-grade AI applications with full diagnostic visibility.

주요 기능

01Standardized span and attribute naming conventions for LLMs and tools

020 GitHub stars

03Cost and token attribution modeling for production monitoring

04Strategic sampling guidelines for high-volume agent systems

055-Tier implementation roadmap from foundation to evaluation

06Multi-agent coordination and handoff tracking patterns

사용 사례

01Standardizing observability across complex multi-agent workflows

02Designing a Day 1 telemetry strategy for a new autonomous agent

03Optimizing agent performance by identifying latency bottlenecks in reasoning steps

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add nexus-labs-automation/agent-observability instrumentation-planning

For use in Claude.ai and ChatGPT

Download Skill