LangChain Reference home pageLangChain ReferenceLangChain Reference
  • GitHub
  • Main Docs
Deep Agents
LangChain
LangGraph
Integrations
LangSmith
LangSmith
  • Client
  • Run Trees
  • Traceable
  • Evaluation
  • Schemas
  • Langchain
  • Jest
  • Vitest
  • Wrappers
  • Anonymizer
  • Jestlike
  • Vercel
  • Anthropic
  • Sandbox
⌘I

LangChain Assistant

Ask a question to get started

Enter to send•Shift+Enter new line

Menu

LangSmith
ClientRun TreesTraceableEvaluationSchemasLangchainJestVitestWrappersAnonymizerJestlikeVercelAnthropicSandbox
Language
Theme
JavaScriptlangsmithclientClient
Class●Since v0.0

Client

Copy
class Client

Used in Docs

  • Call agents from code
  • Collect feedback with presigned URLs
  • Evaluate a RAG application
  • Going to production
  • How to audit evaluator scores
(22 more not shown)

Constructors

Properties

Methods

View source on GitHub
constructor
constructor
property
debug: boolean
property
headers: Record<string, string>
property
tracingMode: TracingMode
method
_currentTenantIsOwner
method
_ensureServerInfo
method
_getLatestCommitHash
method
_getPromptUrl
method
_getServerInfo
method
_getSettings
method
_likeOrUnlikePrompt
method
_logEvaluationFeedback→ Promise<[EvaluationResult[], Feedback[]]>
method
_ownerConflictError
method
_selectEvalResults→ EvaluationResult[]
method
addRunsToAnnotationQueue→ Promise<void>
method
agentExists→ Promise<boolean>
method
awaitPendingTraceBatches→ Promise<void>
method
batchIngestRuns→ Promise<void>
method
cleanup
method
clonePublicDataset→ Promise<void>
method
createAnnotationQueue→ Promise<AnnotationQueueWithDetails>
method
createChatExample→ Promise<Example>
method
createCommit→ Promise<string>
method
createComparativeExperiment→ Promise<ComparativeExperiment>
method
createDataset→ Promise<Dataset>
method
createFeedback→ Promise<Feedback>
method
createFeedbackConfig→ Promise<FeedbackConfigSchema>
method
createLLMExample→ Promise<Example>
method
createPresignedFeedbackToken→ Promise<FeedbackIngestToken>
method
createProject→ Promise<TracerSession>
method
createPrompt→ Promise<Prompt>
method
createRun→ Promise<void>
method
deleteAgent→ Promise<void>
method
deleteAnnotationQueue→ Promise<void>
method
deleteDataset→ Promise<void>
method
deleteExample→ Promise<void>
method
deleteExamples→ Promise<void>
method
deleteFeedback→ Promise<void>
method
deleteFeedbackConfig→ Promise<void>
method
deleteProject→ Promise<void>
method
deletePrompt→ Promise<void>
method
deleteRunFromAnnotationQueue→ Promise<void>
method
deleteSkill→ Promise<void>
method
diffDatasetVersions→ Promise<DatasetDiffInfo>
method
flush→ Promise<void>
method
getDatasetUrl→ Promise<string>
method
getHostUrl→ string
method
getProjectUrl→ Promise<string>
method
getPrompt→ Promise<Prompt | null>
method
getRunFromAnnotationQueue→ Promise<RunWithAnnotationQueueInfo>
method
getRunStats→ Promise<any>
method
getRunUrl→ Promise<string>
method
getSizeFromAnnotationQueue→ Promise<__type>
method
hasDataset→ Promise<boolean>
method
hasProject→ Promise<boolean>
method
likePrompt→ Promise<LikePromptResponse>
method
listAgents→ AsyncIterableIterator<Prompt>
method
listAnnotationQueues→ AsyncIterableIterator<AnnotationQueue>
method
listCommits→ AsyncIterableIterator<PromptCommit>
method
listDatasets→ AsyncIterable<Dataset>
method
listDatasetSplits→ Promise<string[]>
method
listExamples→ AsyncIterable<Example>
method
listFeedback→ AsyncIterable<Feedback>
method
listFeedbackConfigs→ AsyncIterableIterator<FeedbackConfigSchema>
method
listGroupRuns→ AsyncIterable<Thread>
method
listPresignedFeedbackTokens→ AsyncIterable<FeedbackIngestToken>
method
listProjects→ AsyncIterable<TracerSessionResult>
method
listPrompts→ AsyncIterableIterator<Prompt>
method
listRuns→ AsyncIterable<Run>
method
listSharedExamples→ Promise<Example[]>
method
listSharedRuns→ Promise<Run[]>
method
listSkills→ AsyncIterableIterator<Prompt>
method
listThreads→ Promise<ListThreadsItem[]>
method
logEvaluationFeedback→ Promise<EvaluationResult[]>
method
multipartIngestRuns→ Promise<void>
method
promptExists→ Promise<boolean>
method
pullAgent→ Promise<AgentContext>
method
pullPromptCommit→ Promise<PromptCommit>
method
pullSkill→ Promise<SkillContext>
method
pushAgent→ Promise<string>
method
pushPrompt→ Promise<string>
method
pushSkill→ Promise<string>
method
readAnnotationQueue→ Promise<AnnotationQueueWithDetails>
method
readDataset→ Promise<Dataset>
method
readDatasetOpenaiFinetuning→ Promise<any[]>
method
readDatasetSharedSchema→ Promise<DatasetShareSchema>
method
readDatasetVersion→ Promise<DatasetVersion>
method
readExample→ Promise<Example>
method
readFeedback→ Promise<Feedback>
method
readProject→ Promise<TracerSessionResult>
method
readRun→ Promise<Run>
method
readRunSharedLink→ Promise<string | undefined>
method
readSharedDataset→ Promise<Dataset>
method
readThread→ AsyncIterable<Run>
method
shareDataset→ Promise<DatasetShareSchema>
method
shareRun→ Promise<string>
method
skillExists→ Promise<boolean>
method
toString→ string
method
unlikePrompt→ Promise<LikePromptResponse>
method
unshareDataset→ Promise<void>
method
unshareRun→ Promise<void>
method
updateAnnotationQueue→ Promise<void>
method
updateDataset→ Promise<Dataset>
method
updateDatasetSplits→ Promise<void>
method
updateDatasetTag→ Promise<void>
method
updateExamples→ Promise<object>
method
updateExamplesMultipart→ Promise<UpdateExamplesResponse>
method
updateFeedback→ Promise<void>
method
updateFeedbackConfig→ Promise<FeedbackConfigSchema>
method
updateProject→ Promise<TracerSession>
method
updatePrompt→ Promise<Record<string, any>>
method
updateRun→ Promise<void>
method
uploadCsv→ Promise<Dataset>
method
getDefaultClientConfig→ DefaultClientConfig
deprecatedmethod
createExample→ Promise<Example>
deprecatedmethod
createExamples→ Promise<Example[]>
deprecatedmethod
updateExample→ Promise<object>
deprecatedmethod
uploadExamplesMultipart→ Promise<UploadExamplesResponse>

Add runs to an annotation queue with the specified queue ID.

Check if an agent repo exists.

Awaits all pending trace batches. Useful for environments where you need to be sure that all tracing requests finish before execution ends, such as serverless environments.

Batch ingest/upsert multiple runs in the Langsmith system.

Cleanup resources held by the client. Stops the cache's background refresh timer.

Clone a public dataset to your own langsmith tenant. This operation is idempotent. If you already have a dataset with the given name, this function will do nothing.

Create an annotation queue on the LangSmith API.

Create a new commit for an existing prompt.

Create a feedback configuration on the LangSmith API.

This upserts: if an identical config already exists, it returns it. If a conflicting config exists for the same key, a 400 error is raised.

Creates a presigned feedback token and URL.

The token can be used to authorize feedback metrics without needing an API key. This is useful for giving browser-based applications the ability to submit feedback without needing to expose an API key.

Create a new prompt.

Delete an agent and all its owned child file repos.

Delete an annotation queue with the specified queue ID.

Delete multiple examples by ID.

Delete a feedback configuration on the LangSmith API.

Delete a run from an an annotation queue.

Delete a skill and all its owned child file repos.

Flushes current queued traces.

Get a prompt by its identifier.

Get a run from an annotation queue at the specified index.

Get the size of an annotation queue.

Like a prompt.

List agent repos. Yields one at a time, auto-paginating.

List the annotation queues on the LangSmith API.

List all commits for a prompt.

List feedback configurations on the LangSmith API.

Retrieves a list of presigned feedback tokens for a given run ID.

List prompts by filter.

List runs from the LangSmith server.

Get shared examples.

List skill repos. Yields one at a time, auto-paginating.

Batch ingest/upsert multiple runs in the Langsmith system.

Check if a prompt exists.

Pull an agent directory from Hub.

Pull a prompt commit from the LangSmith API.

Public prompts referenced by owner/name cross a trust boundary because the prompt manifest may contain serialized LangChain objects and configuration that affect runtime behavior. For example, a prompt can intentionally configure a model with a custom base URL, headers, model name, or other constructor arguments. These are supported features, but they also mean the prompt contents should be treated as executable configuration rather than plain text.

Set dangerouslyPullPublicPrompt: true only after reviewing and trusting the prompt contents, not merely the publishing account. Prompts from your own or your organization's account can still be unsafe if that account or prompt was compromised.

When pulling a trusted external prompt, prefer pinning to a specific commit rather than following a mutable latest version. Using includeModel: true increases risk and should be avoided for public prompts or prompts outside your own organization.

Pull a skill directory from Hub.

Push an agent to Hub. Creates the repo if missing, patches metadata if provided, then commits the given files.

Push a skill to Hub.

Read an annotation queue with the specified queue ID.

Get dataset version by closest date or exact tag.

Use this to resolve the nearest version to a given timestamp or for a given tag.

Check if a skill repo exists.

Returns a string representation of the Client instance. This method is called when the object is converted to a string or logged, ensuring sensitive information like API keys is not exposed.

Unlike a prompt (remove a previously added like).

Update an annotation queue with the specified queue ID.

Update a dataset

Updates a tag on a dataset.

If the tag is already assigned to a different version of this dataset, the tag will be moved to the new version. The as_of parameter is used to determine which version of the dataset to apply the new tags to.

It must be an exact version of the dataset to succeed. You can use the "readDatasetVersion" method to find the exact version to apply the tags to.

Update examples with attachments using multipart form data.

Update a feedback configuration on the LangSmith API.

Upload examples with attachments using multipart form data.