Debugging Deep Agents with LangSmith

Debugging complex "deep agents" presents unique challenges compared to simpler LLM applications because of their long-running nature, extensive prompts, and multi-turn interactions. LangSmith addresses this by providing detailed tracing capabilities that capture every step of an agent's execution in a structured format. To help developers analyze this data, LangChain has introduced Polly, an in-app AI assistant that can analyze traces and suggest prompt improvements. Additionally, the new langsmith-fetch CLI allows developers to easily retrieve trace data from their terminal for local analysis or to feed into other coding agents.