Tag
LangChain introduces Trajectories in LangSmith, a chronological view of agent sessions that simplifies debugging by aggregating messages from humans, AI, and tools in order for easier navigation and analysis.
LangChain announces new features in LangSmith from their keynote, including LangSmith Engine v2 for red teaming, Managed Deep Agents v0.8 with new auth and memory, LangSmith Trajectories for agent session views, and LangSmith Fine-Tuning.
This paper studies harness design for LLM agents, separating it into task decomposition and guided execution, and shows that more elaborate harnesses are not uniformly better; it reveals failure modes and proposes partial harnesses as effective.