
Trace Every Cursor Agent Turn in LangSmith
LangChain21 July 2026Watch on YouTube
Part of series
Ep. 10 · LangChain Deep Agents
Stap-voor-stap uitleg en tutorials over het bouwen, configureren en deployen van Deep Agents met het LangChain-framework.
View the seriesWhat you'll learn
- Learn how to install and configure the LangSmith plugin in Cursor for complete agent tracing
- Discover how to set three essential environment variables to enable tracing
- See how Cursor agent actions appear as full traces in LangSmith with a structured format
- Understand how attachments are displayed and organized within traces
Frequently asked questions
What are the three environment variables you need to set for Cursor tracing?
How do you verify that the LangSmith integration in Cursor is working correctly?
Can you compare Cursor traces with traces from other AI agents?
What happens after installing the LangSmith plugin in Cursor?
Topics
Description from the channel
Amy from the LangChain product team walks through wiring up Cursor so every agent turn lands in LangSmith as a full trace. You'll install the plugin, set three environment variables, and confirm the setup by running a real Cursor agent task end to end. This is part of a series that also covers Claude Code and Codex, so traces from all three agents show up in a common structure. Chapters: 0:00 What this walkthrough covers 0:13 What you need before you start 0:25 Installing the LangSmith plugin in Cursor 0:45 Confirming the install and restarting Cursor 0:56 Setting your environment variables 1:04 The three variables that power tracing 1:24 Confirming the hook is live 1:31 Running a Cursor agent task 1:53 Opening the trace in LangSmith 1:59 How each trace is structured 2:21 How attachments show up in traces 2:44 Comparing Cursor, Claude Code, and Codex traces Resources: → LangSmith: https://www.langchain.com/langsmith → Trace Claude Code sessions in LangSmith: https://www.youtube.com/watch?v=jLOM_ahG78c → Trace Codex sessions in LangSmith: https://www.youtube.com/watch?v=jv91Ruwrl0U → LangChain Academy: https://academy.langchain.com