Search Results: langsmith-eval

langgraph-testing-evaluation

Use this skill when you need to test or evaluate LangGraph/LangChain agents: writing unit or integration tests, generating test scaffolds, mocking LLM/tool behavior, running trajectory evaluation (match or LLM-as-judge), running LangSmith dataset evaluations, and comparing two agent versions with A/B-style offline analysis. Use it for Python and JavaScript/TypeScript workflows, evaluator design, experiment setup, regression gates, and debugging flaky/incorrect evaluation results.