LangSmith

LangSmith is the developer platform from the makers of LangChain. It traces what happens inside an LLM application, so when an agent does something odd you can open the run and see each step, tool call and model response. You can build test datasets, run evaluations, compare prompt versions and watch production traffic for problems. A prompt playground lets you tweak and re-run a prompt using a real trace as the starting point. It works with LangChain and LangGraph but also with plain code. There is a free developer plan and paid tiers for teams.