Can I use this to debug LangGraph state machines?

Absolutely. This skill is specifically designed to handle the complex, nested traces generated by LangGraph agents as well as standard LangChain implementations.

How does the error detection workflow work?

Claude can fetch a batch of recent traces, grep for failure keywords, and then perform a root-cause analysis on the JSON data to explain exactly why a tool or model call failed.

Does this skill require a LangSmith account?

Yes, you must have a LangSmith API key and an active project. You also need to install the 'langsmith-fetch' CLI via pip.

What is the LangSmith Fetch skill for Claude Code?

It is a specialized capability that allows Claude to interact with the LangSmith API to fetch, analyze, and troubleshoot execution traces from your AI agents directly in the terminal.

Can I export my debug sessions for my team?

Yes, the skill includes a workflow to export traces and conversation threads into timestamped directories, making it easy to share logs or archive them for future reference.

LangSmith Trace Debugger

Name: LangSmith Trace Debugger
Author: ai-wes

byai-wes

분석 및 모니터링

Debugs and analyzes LangChain and LangGraph agent execution traces directly within the terminal using LangSmith Studio data.

소개

LangSmith Trace Debugger is a specialized Claude Code Skill designed to streamline the debugging process for LangChain and LangGraph applications. It allows developers to retrieve real-time execution traces, inspect tool-calling patterns, monitor token usage, and investigate agent failures without leaving the CLI environment. By bridging the gap between local development and LangSmith's observability platform, it enables rapid root-cause analysis of complex agentic workflows, memory operations, and performance bottlenecks.

주요 기능

Fetch and analyze recent LangChain/LangGraph execution traces in real-time
Automated error detection and categorization across multiple agent runs
Monitor performance metrics including execution latency and token costs
Perform deep-dive JSON analysis of specific trace IDs to pinpoint failure nodes
Export full debug sessions and conversation threads for team collaboration
0 GitHub stars

사용 사례

Auditing agent memory operations to ensure long-term context is handled correctly
Optimizing LLM costs and latency by analyzing token usage and execution flow
Identifying why an AI agent failed a specific task or triggered a tool error

소개

주요 기능

Fetch and analyze recent LangChain/LangGraph execution traces in real-time
Automated error detection and categorization across multiple agent runs
Monitor performance metrics including execution latency and token costs
Perform deep-dive JSON analysis of specific trace IDs to pinpoint failure nodes
Export full debug sessions and conversation threads for team collaboration
0 GitHub stars

사용 사례

Auditing agent memory operations to ensure long-term context is handled correctly
Optimizing LLM costs and latency by analyzing token usage and execution flow
Identifying why an AI agent failed a specific task or triggered a tool error