Do I need a LangSmith account to use this skill?

Yes, you must have an active LangSmith account, an API key, and a project name configured in your environment variables.

How do I install the necessary dependencies?

You can install the required CLI tool by running 'pip install langsmith-fetch' in your terminal.

Does it support LangGraph agents?

Absolutely. It is designed to work with any agent architecture that logs to LangSmith, including LangChain and LangGraph.

Can I use this skill to monitor token costs?

Yes, the skill fetches metadata that includes token usage and execution time, helping you track the cost and performance of your AI agents.

Can I export trace data for my team?

Yes, the skill includes an 'Export Debug Session' workflow that saves traces and threads into organized local folders for sharing.

LangSmith Agent Debugger

Name: LangSmith Agent Debugger
Author: Tako3832

byTako3832

0•

Analytics & Monitoring

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

This skill integrates LangSmith's powerful observability features into the Claude Code environment, enabling developers to diagnose agent failures, inspect tool calls, and analyze memory operations without leaving the terminal. It provides a bridge between high-level agent behavior and low-level execution logs, allowing for real-time investigation of why an agent might be failing, hanging, or producing incorrect results, while offering actionable fixes and performance auditing.

Key Features

01Fetch and visualize recent LangChain execution traces in real-time

02Automated error detection and frequency pattern recognition across logs

03Export comprehensive debug sessions for team review and documentation

04Perform deep-dive JSON analysis of specific Trace IDs and tool outputs

05Detailed performance auditing including token usage and tool call latency

060 GitHub stars

Use Cases

01Auditing agent memory operations to verify correct retrieval and storage

02Optimizing agent performance by identifying slow chains and excessive token usage

03Troubleshooting failed tool calls or connection timeouts in complex AI agents

Key Features

01Fetch and visualize recent LangChain execution traces in real-time

02Automated error detection and frequency pattern recognition across logs

03Export comprehensive debug sessions for team review and documentation

04Perform deep-dive JSON analysis of specific Trace IDs and tool outputs

05Detailed performance auditing including token usage and tool call latency

060 GitHub stars

Use Cases

01Auditing agent memory operations to verify correct retrieval and storage

02Optimizing agent performance by identifying slow chains and excessive token usage

03Troubleshooting failed tool calls or connection timeouts in complex AI agents