Can I use this to debug LangGraph state machines?

Yes, it is fully compatible with LangGraph and can help trace state transitions and identify which node in your graph triggered an error.

What is the primary purpose of the LangSmith Fetch skill?

It allows Claude to fetch and analyze execution traces from LangSmith Studio to debug agent behavior, tool calls, and performance issues directly within the Claude Code environment.

How does it help with token optimization?

The skill fetches metadata from traces that include total token counts and costs, allowing you to identify which parts of your agent prompt are consuming the most resources.

Does this skill require external software?

Yes, you must have the langsmith-fetch CLI installed via pip and a valid LangSmith API key configured in your environment variables.

Can I share the debug results with my team?

Yes, the skill includes a workflow to export debug sessions into organized local folders, making it easy to share traces and analysis via Git or other collaboration tools.

LangSmith Fetch

Name: LangSmith Fetch
Author: frostant

byfrostant

•

분석 및 모니터링

Debugs LangChain and LangGraph agents by fetching and analyzing execution traces from LangSmith Studio directly in the terminal.

소개

LangSmith Fetch is a specialized debugging skill that empowers Claude to monitor and troubleshoot agentic workflows built with LangChain and LangGraph. By interfacing with the LangSmith Studio CLI, it provides immediate visibility into execution traces, tool call sequences, memory operations, and token usage. This skill is essential for developers looking to perform root cause analysis on failed runs, optimize agent latency, and verify that complex state machines or long-term memory operations are functioning as intended without switching between the terminal and a browser.

주요 기능

1 GitHub stars
Real-time execution trace retrieval for LangChain and LangGraph agents
Performance monitoring for execution latency and token consumption
Automated root cause analysis for failed agent tool calls and logic
Granular deep-dives into specific Trace IDs and execution threads
Debug session export functionality for local archiving and team collaboration

사용 사례

Troubleshooting agent failures by identifying specific tool call errors or logic gaps
Verifying memory persistence and retrieval accuracy in complex LangGraph agents
Analyzing and optimizing LLM token usage and execution costs during development

소개

주요 기능

1 GitHub stars
Real-time execution trace retrieval for LangChain and LangGraph agents
Performance monitoring for execution latency and token consumption
Automated root cause analysis for failed agent tool calls and logic
Granular deep-dives into specific Trace IDs and execution threads
Debug session export functionality for local archiving and team collaboration

사용 사례

Troubleshooting agent failures by identifying specific tool call errors or logic gaps
Verifying memory persistence and retrieval accuracy in complex LangGraph agents
Analyzing and optimizing LLM token usage and execution costs during development