What problem does it solve? Debugging LangChain and LangGraph agents requires digging through LangSmith Studio manually to find errors, failed tool calls, and performance bottlenecks. This Skill brings trace fetching and analysis directly into the terminal so failures can be diagnosed without leaving the development workflow. ## Core Features & Use Cases - Recent Trace Inspection: Fetch traces from the last N minutes with pretty, JSON, or raw output formats to quickly see what an agent just did. - Deep Trace Analysis: Investigate a specific trace ID to reconstruct execution flow, identify the exact failing tool call, and get root-cause analysis with suggested fixes. - Error Detection & Session Export: Grep recent traces for errors and failure patterns, or export full debug sessions (traces plus threads) into organized timestamped folders for sharing or archiving. - Use Case: An agent silently fails during a Neo4j query. Run a quick trace fetch, spot the connection timeout in the failing tool call, and get a concrete recommendation such as adding retry logic. ## Quick Start Ask the assistant to show what my agent did in the last 5 minutes and flag any errors using langsmith-fetch.
