LangSmith
The official LangSmith MCP server connects AI agents to LangChain's LangSmith platform. Agents can fetch conversation thread history, list and read prompts, query runs and traces, browse projects, datasets, and examples, list and run experiments, and check organization billing usage — for LLM observability and evaluation.
Tools
create_dataset
Learn how to create a new evaluation dataset in LangSmith.
fetch_runs
Fetch LangSmith runs and traces from one or more projects.
get_billing_usage
Fetch organization billing usage for a date range.
get_prompt_by_name
Get a specific LangSmith prompt by its exact name.
get_thread_history
Retrieve the message history for a conversation thread.
list_datasets
Fetch LangSmith datasets filtered by id, type, or name.
list_examples
Fetch examples from a LangSmith dataset with filters.
list_experiments
List experiment projects for a dataset with metrics.
list_projects
List LangSmith projects with optional name filtering.
list_prompts
Fetch prompts from LangSmith filtered by visibility.
push_prompt
Learn how to create and push prompts to LangSmith.
read_dataset
Read a single LangSmith dataset by its id or name.
read_example
Read a single LangSmith example, optionally by version.
run_experiment
Learn how to run experiments and evaluations in LangSmith.
update_examples
Learn how to update the examples in a LangSmith dataset.