opik 2.0.0 → 2.0.2

package/dist/index.d.cts

@@ -11854,21 +11854,52 @@ interface BasePromptData {
  * Provides common functionality for versioning, property updates, and deletion.
  */
 declare abstract class BasePrompt {
-    readonly id: string | undefined;
-    readonly versionId: string | undefined;
-    readonly commit: string | undefined;
+    private _id;
+    private _versionId;
+    private _commit;
+    private _synced;
+    private _changeDescription;
+    private _projectName;
     readonly type: PromptType;
-    readonly changeDescription: string | undefined;
     readonly templateStructure: PromptTemplateStructure;
-    readonly projectName: string | undefined;
-    /** Whether the prompt has been successfully synced with the backend. */
-    readonly synced: boolean;
     protected _name: string;
     protected _description: string | undefined;
     protected _tags: string[];
     protected readonly _metadata: JsonNode | undefined;
     protected readonly opik: OpikClient;
-    constructor(data: BasePromptData, opik: OpikClient);
+    /** Pending background sync promise, set when constructed without synced:true. */
+    protected _pendingSync: Promise<void> | null;
+    get id(): string | undefined;
+    get versionId(): string | undefined;
+    get commit(): string | undefined;
+    /** Whether the prompt has been successfully synced with the backend. */
+    get synced(): boolean;
+    get changeDescription(): string | undefined;
+    get projectName(): string | undefined;
+    constructor(data: BasePromptData, opik?: OpikClient);
+    /**
+     * Updates internal state after a successful background sync.
+     */
+    protected updateSyncState(result: {
+        promptId?: string;
+        versionId?: string;
+        commit?: string;
+        changeDescription?: string;
+        tags?: string[];
+        projectName?: string;
+    }): void;
+    /**
+     * Shared background-sync helper for subclass constructors.
+     * Calls the provided create function, then only updates sync state when the
+     * returned instance is truly synced (has id, versionId, and commit populated).
+     * If the create call throws, or returns an unsynced instance, _synced stays false.
+     */
+    protected _syncViaCreate<T extends BasePrompt>(create: () => Promise<T>): Promise<void>;
+    /**
+     * Resolves when initialization completes.
+     * Returns immediately if the prompt was already persisted (e.g. retrieved from backend).
+     */
+    ready(): Promise<void>;
     get name(): string;
     get description(): string | undefined;
     get tags(): readonly string[] | undefined;
@@ -11924,10 +11955,10 @@ declare abstract class BasePrompt {
      */
     protected retrieveVersionByCommit(commit: string): Promise<PromptVersionDetail | null>;
     /**
-     * Throws an error if the prompt has not been synced with the backend.
-     * Call syncWithBackend() first to sync.
+     * Throws an error if the prompt has not been successfully synced with the backend.
+     * Used internally before backend operations to ensure we have a valid prompt ID.
      */
-    protected requireSynced(operation: string): void;
+    protected ensureSynced(operation: string): void;
     /**
      * Get a specific version by commit hash.
      * Returns a new instance of the appropriate prompt type.
@@ -11966,9 +11997,12 @@ declare class Prompt extends BasePrompt {
     readonly prompt: string;
     /**
      * Creates a new Prompt instance.
-     * This should not be created directly, use OpikClient.createPrompt() instead.
+     * All operations work seamlessly without requiring manual configuration.
      */
+    constructor(data: PromptData);
+    /** @deprecated Passing an opik client is deprecated. */
     constructor(data: PromptData, opik: OpikClient);
+    private _performSync;
     /**
      * Returns the template string for this text prompt.
      * Alias for the `prompt` property for consistency with ChatPrompt.
@@ -12040,23 +12074,6 @@ declare class Prompt extends BasePrompt {
      * ```
      */
     useVersion(version: PromptVersion): Promise<Prompt>;
-    /**
-     * Get a Prompt with a specific version by commit hash.
-     *
-     * @param commit - Commit hash (8-char short form or full)
-     * @returns Prompt instance representing that version, or null if not found
-     *
-     * @example
-     * ```typescript
-     * const prompt = await client.getPrompt({ name: "greeting" });
-     *
-     * // Get a specific version directly as a Prompt
-     * const versionedPrompt = await prompt.getVersion("abc123de");
-     * if (versionedPrompt) {
-     *   const text = versionedPrompt.format({ name: "Alice" });
-     * }
-     * ```
-     */
     /**
      * Synchronize the prompt with the backend.
      *
@@ -12066,6 +12083,12 @@ declare class Prompt extends BasePrompt {
      * @returns Promise resolving to a new synced Prompt instance, or this instance if sync fails
      */
     syncWithBackend(): Promise<Prompt>;
+    /**
+     * Get a Prompt with a specific version by commit hash.
+     *
+     * @param commit - Commit hash (8-char short form or full)
+     * @returns Prompt instance representing that version, or null if not found
+     */
     getVersion(commit: string): Promise<Prompt | null>;
 }
 
@@ -12503,9 +12526,10 @@ declare class Dataset<T extends DatasetItemData = DatasetItemData> {
      * Retrieve raw DatasetItem objects with full metadata (evaluators, executionPolicy) preserved.
      *
      * @param nbSamples The number of samples to retrieve. If not set - all items are returned
+     * @param lastRetrievedId Optional ID of the last retrieved item for pagination
      * @returns A list of DatasetItem objects
      */
-    getRawItems(nbSamples?: number): Promise<DatasetItem<T>[]>;
+    getRawItems(nbSamples?: number, lastRetrievedId?: string): Promise<DatasetItem<T>[]>;
     /**
      * Insert items from a JSON string array into the dataset.
      *
@@ -12763,6 +12787,10 @@ interface EvaluateTestSuiteOptions<T = Record<string, unknown>> {
     client?: OpikClient;
     /** Optional list of tags to associate with the experiment */
     tags?: string[];
+    /** Number of concurrent task executions (default: 16, matching Python SDK) */
+    taskThreads?: number;
+    /** Limit the number of dataset items to evaluate. If not set, all items are evaluated. */
+    nbSamples?: number;
 }
 /**
  * Run a test suite using evaluators and execution policy stored in the dataset version metadata.
@@ -12783,7 +12811,6 @@ interface CreateTestSuiteOptions {
 interface UpdateTestSuiteOptions {
     globalAssertions?: string[];
     globalExecutionPolicy?: ExecutionPolicy;
-    tags?: string[];
 }
 declare class TestSuite {
     private readonly _dataset;
@@ -12820,7 +12847,7 @@ declare class TestSuite {
      * @throws Error if any item is missing an `id`
      */
     update(items: UpdateTestSuiteItem[]): Promise<void>;
-    getItems(): Promise<Array<{
+    getItems(nbSamples?: number, lastRetrievedId?: string): Promise<Array<{
         id: string;
         data: Record<string, unknown>;
         description?: string;
@@ -12889,6 +12916,10 @@ interface RunTestsOptions {
     experimentTags?: string[];
     /** Optional model name override for LLMJudge evaluators */
     model?: string;
+    /** Number of concurrent task executions (default: 16, matching Python SDK) */
+    taskThreads?: number;
+    /** Limit the number of dataset items to evaluate. If not set, all items are evaluated. */
+    nbSamples?: number;
 }
 /**
  * Run a test suite evaluation against a task function.
@@ -13133,9 +13164,12 @@ declare class ChatPrompt extends BasePrompt {
     private readonly chatTemplate;
     /**
      * Creates a new ChatPrompt instance.
-     * This should not be created directly, use OpikClient.createChatPrompt() instead.
+     * All operations work seamlessly without requiring manual configuration.
      */
+    constructor(data: ChatPromptData);
+    /** @deprecated Passing an opik client is deprecated. */
     constructor(data: ChatPromptData, opik: OpikClient);
+    private _performSync;
     /**
      * Returns the template messages for this chat prompt.
      * Alias for the `messages` property for consistency with Prompt.
@@ -13216,23 +13250,6 @@ declare class ChatPrompt extends BasePrompt {
      * ```
      */
     useVersion(version: PromptVersion): Promise<ChatPrompt>;
-    /**
-     * Get a ChatPrompt with a specific version by commit hash.
-     *
-     * @param commit - Commit hash (8-char short form or full)
-     * @returns ChatPrompt instance representing that version, or null if not found
-     *
-     * @example
-     * ```typescript
-     * const chatPrompt = await client.getChatPrompt({ name: "greeting" });
-     *
-     * // Get a specific version directly as a ChatPrompt
-     * const versionedPrompt = await chatPrompt.getVersion("abc123de");
-     * if (versionedPrompt) {
-     *   const messages = versionedPrompt.format({ name: "Alice" });
-     * }
-     * ```
-     */
     /**
      * Synchronize the chat prompt with the backend.
      *
@@ -13242,6 +13259,12 @@ declare class ChatPrompt extends BasePrompt {
      * @returns Promise resolving to a new synced ChatPrompt instance, or this instance if sync fails
      */
     syncWithBackend(): Promise<ChatPrompt>;
+    /**
+     * Get a ChatPrompt with a specific version by commit hash.
+     *
+     * @param commit - Commit hash (8-char short form or full)
+     * @returns ChatPrompt instance representing that version, or null if not found
+     */
     getVersion(commit: string): Promise<ChatPrompt | null>;
 }
 
@@ -13940,6 +13963,11 @@ declare class OpikClient {
      * Throws ConfigMismatchError on the first mismatch found.
      */
     private _validatePromptProjects;
+    /**
+     * Waits for all unsynced BasePrompt values in `values` to finish syncing,
+     * with a timeout. Returns true only when every prompt is synced.
+     */
+    private _allPromptsSynced;
     private _makeFallbackConfig;
     /**
      * Fetches a blueprint from the backend (or returns the cached one).
@@ -14118,6 +14146,8 @@ interface EvaluateOptions<T = Record<string, unknown>> {
     scoringKeyMapping?: ScoringKeyMappingType;
     /** Optional list of tags to associate with the experiment */
     tags?: string[];
+    /** Number of concurrent task executions (default: 16, matching Python SDK) */
+    taskThreads?: number;
     /** Optional agent configuration blueprint ID to link with the experiment */
     blueprintId?: string;
 }
@@ -15851,6 +15881,10 @@ declare class DatasetVersionNotFoundError extends OpikError {
     constructor(versionName: string, datasetName: string);
 }
 
+declare function getGlobalClient(): OpikClient;
+declare function setGlobalClient(client: OpikClient): void;
+declare function resetGlobalClient(): void;
+
 /**
  * Type definitions for OQL (Opik Query Language) parser
  */
@@ -15979,4 +16013,4 @@ declare class ConfigMismatchError extends OpikError {
 
 declare function activateRunner(): void;
 
-export { AgentTaskCompletionJudge, AgentToolCorrectnessJudge, type AllProviderOptions, AnnotationQueuePublicScope as AnnotationQueueScope, AnswerRelevance, type AnthropicProviderOptions, BaseLLMJudgeMetric, BaseMetric, BaseSuiteEvaluator, ChatPrompt, ComplianceRiskJudge, type Config, ConfigMismatchError, ConfigNotFoundError, Contains, type CreateTestSuiteOptions, DEFAULT_EXECUTION_POLICY, Dataset, type DatasetPublic, DatasetVersion, DatasetVersionNotFoundError, type DatasetVersionPublic, DemographicBiasJudge, DialogueHelpfulnessJudge, type ErrorInfo, type EvaluateOptions, type EvaluatePromptOptions, type EvaluateTestSuiteOptions, type EvaluationError, type EvaluationResult, type EvaluationScoreResult, type EvaluationTask, type EvaluationTestCase, type EvaluationTestResult, ExactMatch, type ExecutionPolicy, type FeedbackScoreData, type FewShotExampleAnswerRelevanceNoContext, type FewShotExampleAnswerRelevanceWithContext, type FewShotExampleHallucination, type FewShotExampleModeration, type FilterExpression, GEval, GEvalPreset, GenderBiasJudge, type GoogleProviderOptions, Hallucination, IsJson, type ItemResult, LLMJudge, type LLMJudgeConfig, type LLMJudgeModelSettings, type LLMJudgeOptions, type LLMJudgeResponseFormat, ModelConfigurationError, ModelError, ModelGenerationError, Moderation, type OpenAIProviderOptions, OpikClient as Opik, type OpikAssistantMessage, OpikBaseModel, type OpikConfig, type OpikMessage, OpikQueryLanguage, SpanType as OpikSpanType, type OpikSystemMessage, type OpikToolMessage, type OpikUserMessage, type Param, PoliticalBiasJudge, Prompt, PromptType, PromptUncertaintyJudge, type ProviderOptionsForModel, QARelevanceJudge, RegexMatch, RegionalBiasJudge, type RegistryEntry, ReligiousBiasJudge, ResponseSchema, type RunTestsOptions, SYSTEM_PROMPT, type ScoringKeyMappingType, Span, SpanType, SummarizationCoherenceJudge, SummarizationConsistencyJudge, type SupportedModelId, TestSuite, type TestSuiteItem, TestSuiteResult, ThreadsAnnotationQueue, Trace, TracesAnnotationQueue, USER_PROMPT_TEMPLATE, type UpdateTestSuiteItem, type UpdateTestSuiteOptions, Usefulness, VercelAIChatModel, activateRunner, agentConfigContext, buildSuiteResult, createModel, createModelFromInstance, deserializeEvaluators, detectProvider, disableLogger, evaluate, evaluatePrompt, evaluateTestSuite, flushAll, generateId, getTrackContext, logger, resolveEvaluators, resolveExecutionPolicy, resolveItemExecutionPolicy, resolveModel, runTests, serializeEvaluators, setLoggerLevel, track, validateEvaluators, validateExecutionPolicy };
+export { AgentTaskCompletionJudge, AgentToolCorrectnessJudge, type AllProviderOptions, AnnotationQueuePublicScope as AnnotationQueueScope, AnswerRelevance, type AnthropicProviderOptions, BaseLLMJudgeMetric, BaseMetric, BaseSuiteEvaluator, ChatPrompt, ComplianceRiskJudge, type Config, ConfigMismatchError, ConfigNotFoundError, Contains, type CreateTestSuiteOptions, DEFAULT_EXECUTION_POLICY, Dataset, type DatasetPublic, DatasetVersion, DatasetVersionNotFoundError, type DatasetVersionPublic, DemographicBiasJudge, DialogueHelpfulnessJudge, type ErrorInfo, type EvaluateOptions, type EvaluatePromptOptions, type EvaluateTestSuiteOptions, type EvaluationError, type EvaluationResult, type EvaluationScoreResult, type EvaluationTask, type EvaluationTestCase, type EvaluationTestResult, ExactMatch, type ExecutionPolicy, type FeedbackScoreData, type FewShotExampleAnswerRelevanceNoContext, type FewShotExampleAnswerRelevanceWithContext, type FewShotExampleHallucination, type FewShotExampleModeration, type FilterExpression, GEval, GEvalPreset, GenderBiasJudge, type GoogleProviderOptions, Hallucination, IsJson, type ItemResult, LLMJudge, type LLMJudgeConfig, type LLMJudgeModelSettings, type LLMJudgeOptions, type LLMJudgeResponseFormat, ModelConfigurationError, ModelError, ModelGenerationError, Moderation, type OpenAIProviderOptions, OpikClient as Opik, type OpikAssistantMessage, OpikBaseModel, type OpikConfig, type OpikMessage, OpikQueryLanguage, SpanType as OpikSpanType, type OpikSystemMessage, type OpikToolMessage, type OpikUserMessage, type Param, PoliticalBiasJudge, Prompt, PromptType, PromptUncertaintyJudge, type ProviderOptionsForModel, QARelevanceJudge, RegexMatch, RegionalBiasJudge, type RegistryEntry, ReligiousBiasJudge, ResponseSchema, type RunTestsOptions, SYSTEM_PROMPT, type ScoringKeyMappingType, Span, SpanType, SummarizationCoherenceJudge, SummarizationConsistencyJudge, type SupportedModelId, TestSuite, type TestSuiteItem, TestSuiteResult, ThreadsAnnotationQueue, Trace, TracesAnnotationQueue, USER_PROMPT_TEMPLATE, type UpdateTestSuiteItem, type UpdateTestSuiteOptions, Usefulness, VercelAIChatModel, activateRunner, agentConfigContext, buildSuiteResult, createModel, createModelFromInstance, deserializeEvaluators, detectProvider, disableLogger, evaluate, evaluatePrompt, evaluateTestSuite, flushAll, generateId, getGlobalClient, getTrackContext, logger, resetGlobalClient, resolveEvaluators, resolveExecutionPolicy, resolveItemExecutionPolicy, resolveModel, runTests, serializeEvaluators, setGlobalClient, setLoggerLevel, track, validateEvaluators, validateExecutionPolicy };

package/dist/index.d.ts

@@ -11854,21 +11854,52 @@ interface BasePromptData {
  * Provides common functionality for versioning, property updates, and deletion.
  */
 declare abstract class BasePrompt {
-    readonly id: string | undefined;
-    readonly versionId: string | undefined;
-    readonly commit: string | undefined;
+    private _id;
+    private _versionId;
+    private _commit;
+    private _synced;
+    private _changeDescription;
+    private _projectName;
     readonly type: PromptType;
-    readonly changeDescription: string | undefined;
     readonly templateStructure: PromptTemplateStructure;
-    readonly projectName: string | undefined;
-    /** Whether the prompt has been successfully synced with the backend. */
-    readonly synced: boolean;
     protected _name: string;
     protected _description: string | undefined;
     protected _tags: string[];
     protected readonly _metadata: JsonNode | undefined;
     protected readonly opik: OpikClient;
-    constructor(data: BasePromptData, opik: OpikClient);
+    /** Pending background sync promise, set when constructed without synced:true. */
+    protected _pendingSync: Promise<void> | null;
+    get id(): string | undefined;
+    get versionId(): string | undefined;
+    get commit(): string | undefined;
+    /** Whether the prompt has been successfully synced with the backend. */
+    get synced(): boolean;
+    get changeDescription(): string | undefined;
+    get projectName(): string | undefined;
+    constructor(data: BasePromptData, opik?: OpikClient);
+    /**
+     * Updates internal state after a successful background sync.
+     */
+    protected updateSyncState(result: {
+        promptId?: string;
+        versionId?: string;
+        commit?: string;
+        changeDescription?: string;
+        tags?: string[];
+        projectName?: string;
+    }): void;
+    /**
+     * Shared background-sync helper for subclass constructors.
+     * Calls the provided create function, then only updates sync state when the
+     * returned instance is truly synced (has id, versionId, and commit populated).
+     * If the create call throws, or returns an unsynced instance, _synced stays false.
+     */
+    protected _syncViaCreate<T extends BasePrompt>(create: () => Promise<T>): Promise<void>;
+    /**
+     * Resolves when initialization completes.
+     * Returns immediately if the prompt was already persisted (e.g. retrieved from backend).
+     */
+    ready(): Promise<void>;
     get name(): string;
     get description(): string | undefined;
     get tags(): readonly string[] | undefined;
@@ -11924,10 +11955,10 @@ declare abstract class BasePrompt {
      */
     protected retrieveVersionByCommit(commit: string): Promise<PromptVersionDetail | null>;
     /**
-     * Throws an error if the prompt has not been synced with the backend.
-     * Call syncWithBackend() first to sync.
+     * Throws an error if the prompt has not been successfully synced with the backend.
+     * Used internally before backend operations to ensure we have a valid prompt ID.
      */
-    protected requireSynced(operation: string): void;
+    protected ensureSynced(operation: string): void;
     /**
      * Get a specific version by commit hash.
      * Returns a new instance of the appropriate prompt type.
@@ -11966,9 +11997,12 @@ declare class Prompt extends BasePrompt {
     readonly prompt: string;
     /**
      * Creates a new Prompt instance.
-     * This should not be created directly, use OpikClient.createPrompt() instead.
+     * All operations work seamlessly without requiring manual configuration.
      */
+    constructor(data: PromptData);
+    /** @deprecated Passing an opik client is deprecated. */
     constructor(data: PromptData, opik: OpikClient);
+    private _performSync;
     /**
      * Returns the template string for this text prompt.
      * Alias for the `prompt` property for consistency with ChatPrompt.
@@ -12040,23 +12074,6 @@ declare class Prompt extends BasePrompt {
      * ```
      */
     useVersion(version: PromptVersion): Promise<Prompt>;
-    /**
-     * Get a Prompt with a specific version by commit hash.
-     *
-     * @param commit - Commit hash (8-char short form or full)
-     * @returns Prompt instance representing that version, or null if not found
-     *
-     * @example
-     * ```typescript
-     * const prompt = await client.getPrompt({ name: "greeting" });
-     *
-     * // Get a specific version directly as a Prompt
-     * const versionedPrompt = await prompt.getVersion("abc123de");
-     * if (versionedPrompt) {
-     *   const text = versionedPrompt.format({ name: "Alice" });
-     * }
-     * ```
-     */
     /**
      * Synchronize the prompt with the backend.
      *
@@ -12066,6 +12083,12 @@ declare class Prompt extends BasePrompt {
      * @returns Promise resolving to a new synced Prompt instance, or this instance if sync fails
      */
     syncWithBackend(): Promise<Prompt>;
+    /**
+     * Get a Prompt with a specific version by commit hash.
+     *
+     * @param commit - Commit hash (8-char short form or full)
+     * @returns Prompt instance representing that version, or null if not found
+     */
     getVersion(commit: string): Promise<Prompt | null>;
 }
 
@@ -12503,9 +12526,10 @@ declare class Dataset<T extends DatasetItemData = DatasetItemData> {
      * Retrieve raw DatasetItem objects with full metadata (evaluators, executionPolicy) preserved.
      *
      * @param nbSamples The number of samples to retrieve. If not set - all items are returned
+     * @param lastRetrievedId Optional ID of the last retrieved item for pagination
      * @returns A list of DatasetItem objects
      */
-    getRawItems(nbSamples?: number): Promise<DatasetItem<T>[]>;
+    getRawItems(nbSamples?: number, lastRetrievedId?: string): Promise<DatasetItem<T>[]>;
     /**
      * Insert items from a JSON string array into the dataset.
      *
@@ -12763,6 +12787,10 @@ interface EvaluateTestSuiteOptions<T = Record<string, unknown>> {
     client?: OpikClient;
     /** Optional list of tags to associate with the experiment */
     tags?: string[];
+    /** Number of concurrent task executions (default: 16, matching Python SDK) */
+    taskThreads?: number;
+    /** Limit the number of dataset items to evaluate. If not set, all items are evaluated. */
+    nbSamples?: number;
 }
 /**
  * Run a test suite using evaluators and execution policy stored in the dataset version metadata.
@@ -12783,7 +12811,6 @@ interface CreateTestSuiteOptions {
 interface UpdateTestSuiteOptions {
     globalAssertions?: string[];
     globalExecutionPolicy?: ExecutionPolicy;
-    tags?: string[];
 }
 declare class TestSuite {
     private readonly _dataset;
@@ -12820,7 +12847,7 @@ declare class TestSuite {
      * @throws Error if any item is missing an `id`
      */
     update(items: UpdateTestSuiteItem[]): Promise<void>;
-    getItems(): Promise<Array<{
+    getItems(nbSamples?: number, lastRetrievedId?: string): Promise<Array<{
         id: string;
         data: Record<string, unknown>;
         description?: string;
@@ -12889,6 +12916,10 @@ interface RunTestsOptions {
     experimentTags?: string[];
     /** Optional model name override for LLMJudge evaluators */
     model?: string;
+    /** Number of concurrent task executions (default: 16, matching Python SDK) */
+    taskThreads?: number;
+    /** Limit the number of dataset items to evaluate. If not set, all items are evaluated. */
+    nbSamples?: number;
 }
 /**
  * Run a test suite evaluation against a task function.
@@ -13133,9 +13164,12 @@ declare class ChatPrompt extends BasePrompt {
     private readonly chatTemplate;
     /**
      * Creates a new ChatPrompt instance.
-     * This should not be created directly, use OpikClient.createChatPrompt() instead.
+     * All operations work seamlessly without requiring manual configuration.
      */
+    constructor(data: ChatPromptData);
+    /** @deprecated Passing an opik client is deprecated. */
     constructor(data: ChatPromptData, opik: OpikClient);
+    private _performSync;
     /**
      * Returns the template messages for this chat prompt.
      * Alias for the `messages` property for consistency with Prompt.
@@ -13216,23 +13250,6 @@ declare class ChatPrompt extends BasePrompt {
      * ```
      */
     useVersion(version: PromptVersion): Promise<ChatPrompt>;
-    /**
-     * Get a ChatPrompt with a specific version by commit hash.
-     *
-     * @param commit - Commit hash (8-char short form or full)
-     * @returns ChatPrompt instance representing that version, or null if not found
-     *
-     * @example
-     * ```typescript
-     * const chatPrompt = await client.getChatPrompt({ name: "greeting" });
-     *
-     * // Get a specific version directly as a ChatPrompt
-     * const versionedPrompt = await chatPrompt.getVersion("abc123de");
-     * if (versionedPrompt) {
-     *   const messages = versionedPrompt.format({ name: "Alice" });
-     * }
-     * ```
-     */
     /**
      * Synchronize the chat prompt with the backend.
      *
@@ -13242,6 +13259,12 @@ declare class ChatPrompt extends BasePrompt {
      * @returns Promise resolving to a new synced ChatPrompt instance, or this instance if sync fails
      */
     syncWithBackend(): Promise<ChatPrompt>;
+    /**
+     * Get a ChatPrompt with a specific version by commit hash.
+     *
+     * @param commit - Commit hash (8-char short form or full)
+     * @returns ChatPrompt instance representing that version, or null if not found
+     */
     getVersion(commit: string): Promise<ChatPrompt | null>;
 }
 
@@ -13940,6 +13963,11 @@ declare class OpikClient {
      * Throws ConfigMismatchError on the first mismatch found.
      */
     private _validatePromptProjects;
+    /**
+     * Waits for all unsynced BasePrompt values in `values` to finish syncing,
+     * with a timeout. Returns true only when every prompt is synced.
+     */
+    private _allPromptsSynced;
     private _makeFallbackConfig;
     /**
      * Fetches a blueprint from the backend (or returns the cached one).
@@ -14118,6 +14146,8 @@ interface EvaluateOptions<T = Record<string, unknown>> {
     scoringKeyMapping?: ScoringKeyMappingType;
     /** Optional list of tags to associate with the experiment */
     tags?: string[];
+    /** Number of concurrent task executions (default: 16, matching Python SDK) */
+    taskThreads?: number;
     /** Optional agent configuration blueprint ID to link with the experiment */
     blueprintId?: string;
 }
@@ -15851,6 +15881,10 @@ declare class DatasetVersionNotFoundError extends OpikError {
     constructor(versionName: string, datasetName: string);
 }
 
+declare function getGlobalClient(): OpikClient;
+declare function setGlobalClient(client: OpikClient): void;
+declare function resetGlobalClient(): void;
+
 /**
  * Type definitions for OQL (Opik Query Language) parser
  */
@@ -15979,4 +16013,4 @@ declare class ConfigMismatchError extends OpikError {
 
 declare function activateRunner(): void;
 
-export { AgentTaskCompletionJudge, AgentToolCorrectnessJudge, type AllProviderOptions, AnnotationQueuePublicScope as AnnotationQueueScope, AnswerRelevance, type AnthropicProviderOptions, BaseLLMJudgeMetric, BaseMetric, BaseSuiteEvaluator, ChatPrompt, ComplianceRiskJudge, type Config, ConfigMismatchError, ConfigNotFoundError, Contains, type CreateTestSuiteOptions, DEFAULT_EXECUTION_POLICY, Dataset, type DatasetPublic, DatasetVersion, DatasetVersionNotFoundError, type DatasetVersionPublic, DemographicBiasJudge, DialogueHelpfulnessJudge, type ErrorInfo, type EvaluateOptions, type EvaluatePromptOptions, type EvaluateTestSuiteOptions, type EvaluationError, type EvaluationResult, type EvaluationScoreResult, type EvaluationTask, type EvaluationTestCase, type EvaluationTestResult, ExactMatch, type ExecutionPolicy, type FeedbackScoreData, type FewShotExampleAnswerRelevanceNoContext, type FewShotExampleAnswerRelevanceWithContext, type FewShotExampleHallucination, type FewShotExampleModeration, type FilterExpression, GEval, GEvalPreset, GenderBiasJudge, type GoogleProviderOptions, Hallucination, IsJson, type ItemResult, LLMJudge, type LLMJudgeConfig, type LLMJudgeModelSettings, type LLMJudgeOptions, type LLMJudgeResponseFormat, ModelConfigurationError, ModelError, ModelGenerationError, Moderation, type OpenAIProviderOptions, OpikClient as Opik, type OpikAssistantMessage, OpikBaseModel, type OpikConfig, type OpikMessage, OpikQueryLanguage, SpanType as OpikSpanType, type OpikSystemMessage, type OpikToolMessage, type OpikUserMessage, type Param, PoliticalBiasJudge, Prompt, PromptType, PromptUncertaintyJudge, type ProviderOptionsForModel, QARelevanceJudge, RegexMatch, RegionalBiasJudge, type RegistryEntry, ReligiousBiasJudge, ResponseSchema, type RunTestsOptions, SYSTEM_PROMPT, type ScoringKeyMappingType, Span, SpanType, SummarizationCoherenceJudge, SummarizationConsistencyJudge, type SupportedModelId, TestSuite, type TestSuiteItem, TestSuiteResult, ThreadsAnnotationQueue, Trace, TracesAnnotationQueue, USER_PROMPT_TEMPLATE, type UpdateTestSuiteItem, type UpdateTestSuiteOptions, Usefulness, VercelAIChatModel, activateRunner, agentConfigContext, buildSuiteResult, createModel, createModelFromInstance, deserializeEvaluators, detectProvider, disableLogger, evaluate, evaluatePrompt, evaluateTestSuite, flushAll, generateId, getTrackContext, logger, resolveEvaluators, resolveExecutionPolicy, resolveItemExecutionPolicy, resolveModel, runTests, serializeEvaluators, setLoggerLevel, track, validateEvaluators, validateExecutionPolicy };
+export { AgentTaskCompletionJudge, AgentToolCorrectnessJudge, type AllProviderOptions, AnnotationQueuePublicScope as AnnotationQueueScope, AnswerRelevance, type AnthropicProviderOptions, BaseLLMJudgeMetric, BaseMetric, BaseSuiteEvaluator, ChatPrompt, ComplianceRiskJudge, type Config, ConfigMismatchError, ConfigNotFoundError, Contains, type CreateTestSuiteOptions, DEFAULT_EXECUTION_POLICY, Dataset, type DatasetPublic, DatasetVersion, DatasetVersionNotFoundError, type DatasetVersionPublic, DemographicBiasJudge, DialogueHelpfulnessJudge, type ErrorInfo, type EvaluateOptions, type EvaluatePromptOptions, type EvaluateTestSuiteOptions, type EvaluationError, type EvaluationResult, type EvaluationScoreResult, type EvaluationTask, type EvaluationTestCase, type EvaluationTestResult, ExactMatch, type ExecutionPolicy, type FeedbackScoreData, type FewShotExampleAnswerRelevanceNoContext, type FewShotExampleAnswerRelevanceWithContext, type FewShotExampleHallucination, type FewShotExampleModeration, type FilterExpression, GEval, GEvalPreset, GenderBiasJudge, type GoogleProviderOptions, Hallucination, IsJson, type ItemResult, LLMJudge, type LLMJudgeConfig, type LLMJudgeModelSettings, type LLMJudgeOptions, type LLMJudgeResponseFormat, ModelConfigurationError, ModelError, ModelGenerationError, Moderation, type OpenAIProviderOptions, OpikClient as Opik, type OpikAssistantMessage, OpikBaseModel, type OpikConfig, type OpikMessage, OpikQueryLanguage, SpanType as OpikSpanType, type OpikSystemMessage, type OpikToolMessage, type OpikUserMessage, type Param, PoliticalBiasJudge, Prompt, PromptType, PromptUncertaintyJudge, type ProviderOptionsForModel, QARelevanceJudge, RegexMatch, RegionalBiasJudge, type RegistryEntry, ReligiousBiasJudge, ResponseSchema, type RunTestsOptions, SYSTEM_PROMPT, type ScoringKeyMappingType, Span, SpanType, SummarizationCoherenceJudge, SummarizationConsistencyJudge, type SupportedModelId, TestSuite, type TestSuiteItem, TestSuiteResult, ThreadsAnnotationQueue, Trace, TracesAnnotationQueue, USER_PROMPT_TEMPLATE, type UpdateTestSuiteItem, type UpdateTestSuiteOptions, Usefulness, VercelAIChatModel, activateRunner, agentConfigContext, buildSuiteResult, createModel, createModelFromInstance, deserializeEvaluators, detectProvider, disableLogger, evaluate, evaluatePrompt, evaluateTestSuite, flushAll, generateId, getGlobalClient, getTrackContext, logger, resetGlobalClient, resolveEvaluators, resolveExecutionPolicy, resolveItemExecutionPolicy, resolveModel, runTests, serializeEvaluators, setGlobalClient, setLoggerLevel, track, validateEvaluators, validateExecutionPolicy };

package/dist/index.js

@@ -1 +1 @@
-export{oa as AgentTaskCompletionJudge,na as AgentToolCorrectnessJudge,ba as AnswerRelevance,Z as BaseLLMJudgeMetric,y as BaseMetric,z as BaseSuiteEvaluator,m as ChatPrompt,qa as ComplianceRiskJudge,i as ConfigMismatchError,h as ConfigNotFoundError,W as Contains,v as DEFAULT_EXECUTION_POLICY,j as Dataset,f as DatasetVersion,g as DatasetVersionNotFoundError,ia as DemographicBiasJudge,ga as DialogueHelpfulnessJudge,V as ExactMatch,ca as GEval,da as GEvalPreset,ka as GenderBiasJudge,aa as Hallucination,Y as IsJson,M as LLMJudge,D as ModelConfigurationError,B as ModelError,C as ModelGenerationError,_ as Moderation,va as Opik,A as OpikBaseModel,n as OpikQueryLanguage,d as OpikSpanType,ja as PoliticalBiasJudge,l as Prompt,k as PromptType,pa as PromptUncertaintyJudge,ha as QARelevanceJudge,X as RegexMatch,ma as RegionalBiasJudge,la as ReligiousBiasJudge,L as ResponseSchema,J as SYSTEM_PROMPT,fa as SummarizationCoherenceJudge,ea as SummarizationConsistencyJudge,ua as TestSuite,w as TestSuiteResult,p as ThreadsAnnotationQueue,o as TracesAnnotationQueue,K as USER_PROMPT_TEMPLATE,$ as Usefulness,F as VercelAIChatModel,s as activateRunner,q as agentConfigContext,x as buildSuiteResult,G as createModel,H as createModelFromInstance,O as deserializeEvaluators,E as detectProvider,c as disableLogger,T as evaluate,U as evaluatePrompt,R as evaluateTestSuite,r as flushAll,e as generateId,t as getTrackContext,a as logger,ra as resolveEvaluators,P as resolveExecutionPolicy,Q as resolveItemExecutionPolicy,I as resolveModel,S as runTests,N as serializeEvaluators,b as setLoggerLevel,u as track,sa as validateEvaluators,ta as validateExecutionPolicy,wa as z}from'./chunk-2AOEXUQ4.js';
+import {Ba}from'./chunk-FTTQJ4TO.js';export{ra as AgentTaskCompletionJudge,qa as AgentToolCorrectnessJudge,ea as AnswerRelevance,aa as BaseLLMJudgeMetric,B as BaseMetric,C as BaseSuiteEvaluator,p as ChatPrompt,ta as ComplianceRiskJudge,i as ConfigMismatchError,h as ConfigNotFoundError,Z as Contains,y as DEFAULT_EXECUTION_POLICY,j as Dataset,f as DatasetVersion,g as DatasetVersionNotFoundError,la as DemographicBiasJudge,ja as DialogueHelpfulnessJudge,Y as ExactMatch,fa as GEval,ga as GEvalPreset,na as GenderBiasJudge,da as Hallucination,$ as IsJson,P as LLMJudge,G as ModelConfigurationError,E as ModelError,F as ModelGenerationError,ba as Moderation,za as Opik,D as OpikBaseModel,q as OpikQueryLanguage,d as OpikSpanType,ma as PoliticalBiasJudge,o as Prompt,k as PromptType,sa as PromptUncertaintyJudge,ka as QARelevanceJudge,_ as RegexMatch,pa as RegionalBiasJudge,oa as ReligiousBiasJudge,O as ResponseSchema,M as SYSTEM_PROMPT,ia as SummarizationCoherenceJudge,ha as SummarizationConsistencyJudge,xa as TestSuite,z as TestSuiteResult,s as ThreadsAnnotationQueue,r as TracesAnnotationQueue,N as USER_PROMPT_TEMPLATE,ca as Usefulness,I as VercelAIChatModel,v as activateRunner,t as agentConfigContext,A as buildSuiteResult,J as createModel,K as createModelFromInstance,R as deserializeEvaluators,H as detectProvider,c as disableLogger,W as evaluate,X as evaluatePrompt,U as evaluateTestSuite,u as flushAll,e as generateId,l as getGlobalClient,w as getTrackContext,a as logger,n as resetGlobalClient,ua as resolveEvaluators,S as resolveExecutionPolicy,T as resolveItemExecutionPolicy,L as resolveModel,V as runTests,Q as serializeEvaluators,m as setGlobalClient,b as setLoggerLevel,x as track,va as validateEvaluators,wa as validateExecutionPolicy,Aa as z}from'./chunk-FTTQJ4TO.js';Ba();

package/dist/suite-2DEGP66I.js

@@ -0,0 +1 @@
+import {ya}from'./chunk-FTTQJ4TO.js';export{y as DEFAULT_EXECUTION_POLICY,xa as TestSuite,z as TestSuiteResult,A as buildSuiteResult,R as deserializeEvaluators,U as evaluateTestSuite,S as resolveExecutionPolicy,T as resolveItemExecutionPolicy,V as runTests,Q as serializeEvaluators}from'./chunk-FTTQJ4TO.js';ya();

package/package.json

@@ -1,7 +1,7 @@
 {
   "name": "opik",
   "description": "Opik TypeScript and JavaScript SDK",
-  "version": "2.0.0",
+  "version": "2.0.2",
   "repository": {
     "type": "git",
     "url": "git+https://github.com/comet-ml/opik.git",