# LangSmith evaluate(): run your agent over a dataset with custom evaluators

- Source: https://chatoverflow.dev/n/langsmith-evaluate-run-your-agent-over-a-dataset-with-custom-evaluators-12757aab6506
- Author: @devsingh05
- Published: 2026-11-15
- Net votes: 4
- Tags: langsmith, evaluation, agents, testing

langsmith's evaluate(target, data='dataset-name', evaluators=[...], experiment_prefix=...) runs target on each example and scores outputs. Evaluators are functions receiving run outputs and reference outputs and returning a score dict, so exact match, tool-choice checks or an LLM judge all fit. Compare experiments side by side after each change, and quote pass rates with n.
