debug-agent-from-traces

Analyze observability traces to diagnose LLM agent issues.

29|8|Updated Jul 5, 2026
One-click install
npx skills add https://github.com/ContextJet-ai/awesome-llm-observability --skill debug-agent-from-traces
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: debug-agent-from-traces
Source: https://github.com/ContextJet-ai/awesome-llm-observability/tree/main/skills/debug-agent-from-traces
Command: npx skills add https://github.com/ContextJet-ai/awesome-llm-observability --skill debug-agent-from-traces

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) components.

What problem does it solve?

This Skill helps diagnose issues with LLM agents by analyzing their observability traces, enabling users to identify the root cause of problems like incorrect results, empty outputs, slow performance, or high costs.

Core Features & Use Cases

  • Trace Analysis: Analyze observability traces to diagnose issues with LLM agents.
  • Symptom Triage: Offers guidance on where to look first based on symptoms like wrong answers, empty outputs, slow performance, or high costs.
  • Systematic Walk: Provides a systematic approach to walk through the trace tree and identify the divergence point.
  • Common Root Causes: Identifies common root causes of issues, such as wrong/empty retrieved context, prompt/template rendering bugs, swallowed exceptions, retry storms, and context bloat.
  • Regression Testing: Encourages turning fixes into regression tests to prevent future issues.

Quick Start

Analyze the trace of your LLM agent to identify the cause of a slow response by running the debug-agent-from-traces skill.

Frequently Asked Questions about debug-agent-from-traces

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I diagnose slow performance in my LLM agent using observability traces?

Diagnose slow LLM agent performance by analyzing observability traces to systematically walk through the trace tree and identify the exact divergence point causing latency. It helps pinpoint common root causes like retry storms or context bloat.

Why does my LLM agent return empty outputs or incorrect results?

LLM agents return empty outputs or incorrect results due to root causes like wrong retrieved context, prompt rendering bugs, or swallowed exceptions. Trace analysis helps triage these symptoms by inspecting the observability trace tree.

What is trace analysis for LLM agent diagnosis and when do I need it?

Trace analysis for LLM agent diagnosis is the process of inspecting observability trace trees to find why agents fail or underperform. You need it when experiencing high costs, slow performance, or incorrect results and require root cause identification.

Can I use trace analysis to find the root cause of high costs in my LLM agent?

Yes, you can use trace analysis to find the root cause of high LLM agent costs. By walking through the trace tree, you can identify cost drivers like context bloat or retry storms that inflate token usage and operational expenses.

What's the best way to triage LLM agent issues based on symptoms like wrong answers or high costs?

The best way to triage LLM agent issues is by matching symptoms like wrong answers or high costs to guidance on where to look first in the observability trace. This systematic approach helps identify divergence points and common root causes efficiently.

How do I prevent recurring LLM agent failures after diagnosing the root cause?

Prevent recurring LLM agent failures by turning your trace diagnosis fixes into regression tests. After identifying the root cause in the observability trace, implementing regression tests ensures that issues like swallowed exceptions or context bloat do not reappear.