Can I export my debugging results for my team?

Yes, the skill includes workflows to export traces and threads to a local directory, making it easy to share session data or archive it for future reference.

What is the primary purpose of the langsmith-fetch skill?

It allows Claude to interact with the LangSmith CLI to fetch, filter, and analyze agent execution data, helping developers debug LangChain and LangGraph applications.

Can this skill help optimize my LLM spending?

Yes, by fetching metadata from traces, the skill can report token usage and costs, allowing you to identify which agents or tools are most expensive.

Do I need a LangSmith account to use this skill?

Yes, you must have an active LangSmith account, an API key, and the langsmith-fetch CLI installed on your machine.

Does it support complex workflows like LangGraph?

Absolutely. It is specifically designed to handle the complex execution flows and tool calls typical of both LangChain and LangGraph agents.

LangSmith Fetch Debugger

Name: LangSmith Fetch Debugger
Author: nordeim

bynordeim

Analytics & Monitoring

Debugs and analyzes LangChain or LangGraph agent execution traces directly from LangSmith Studio within your terminal.

About

The LangSmith Fetch skill empowers developers to troubleshoot complex AI agent behaviors by fetching and analyzing execution traces without leaving the terminal. It provides deep visibility into tool calls, memory operations, error logs, and token usage, enabling rapid root-cause analysis for failing agents or performance bottlenecks. By automating the retrieval of recent traces and providing structured reports, it simplifies the transition between local development and cloud-based observability.

Key Features

Fetch and analyze recent execution traces with customizable time-based filtering
Monitor performance metrics including execution latency and token consumption
Perform deep-dive investigations of specific trace IDs to identify failure points
0 GitHub stars
Export full debug sessions to organized local directories for collaborative review
Detect common error patterns and pinpoint malfunctioning tools or memory stores

Use Cases

Analyzing why an agent selected a specific tool or failed to retrieve relevant memories
Troubleshooting agent failures or 'hanging' states by inspecting live trace flows
Benchmarking agent performance and token costs across multiple execution iterations

About

Key Features

Fetch and analyze recent execution traces with customizable time-based filtering
Monitor performance metrics including execution latency and token consumption
Perform deep-dive investigations of specific trace IDs to identify failure points
0 GitHub stars
Export full debug sessions to organized local directories for collaborative review
Detect common error patterns and pinpoint malfunctioning tools or memory stores

Use Cases

Analyzing why an agent selected a specific tool or failed to retrieve relevant memories
Troubleshooting agent failures or 'hanging' states by inspecting live trace flows
Benchmarking agent performance and token costs across multiple execution iterations