@anvia/core 0.26.0 → 1.0.0-rc.1

package/dist/embeddings/index.d.ts

@@ -1,5 +1,5 @@
-import { V as VectorMetadata, E as EmbeddingModel, b as EmbedDocumentsOptions, c as EmbeddedDocument, d as EmbedHybridDocumentsOptions, S as SparseEmbeddingModel, e as SparseEmbedding, f as Embedding } from '../types-BCTRUGex.js';
-export { g as SparseVector, a as VectorMetadataValue } from '../types-BCTRUGex.js';
+import { V as VectorMetadata, E as EmbeddingModel, c as EmbedDocumentsOptions, a as EmbeddedDocument, d as EmbedHybridDocumentsOptions, S as SparseEmbeddingModel, e as SparseEmbedding, f as Embedding } from '../types-p55Jc-sr.js';
+export { g as SparseVector, b as VectorMetadataValue } from '../types-p55Jc-sr.js';
 
 declare function dotProduct(left: number[], right: number[]): number;
 declare function cosineSimilarity(left: number[], right: number[]): number;

package/dist/embeddings/index.js

@@ -11,8 +11,7 @@ import {
   embedTexts,
   euclideanDistance,
   manhattanDistance
-} from "../chunk-MNWK2USR.js";
-import "../chunk-OIMLU4SF.js";
+} from "../chunk-DZ3CKYTH.js";
 export {
   angularDistance,
   chebyshevDistance,

package/dist/{errors-2moLDuuN.d.ts → errors-CYKUcU89.d.ts}

@@ -1,5 +1,5 @@
-import { C as CompletionModel, M as Message, U as Usage } from './types-G12i56s2.js';
-import { S as SummaryMemoryCompactorOptions, e as MemoryCompactor } from './types-BrLbKnEz.js';
+import { C as CompletionModel, M as Message, U as Usage } from './types-B3A4YxaA.js';
+import { S as SummaryMemoryCompactorOptions, e as MemoryCompactor } from './types-BBEGKB9m.js';
 
 declare function createSummaryMemoryCompactor(model: CompletionModel, options?: SummaryMemoryCompactorOptions): MemoryCompactor;
 declare function isMemoryCompactionSummary(message: Message): boolean;

package/dist/evals/index.d.ts

@@ -1,15 +1,22 @@
+import { U as Usage, l as JsonValue, C as CompletionModel, M as Message } from '../types-B3A4YxaA.js';
 import { Z as ZodSchema } from '../zod-schema-C7F4clpm.js';
-import { U as Usage, l as JsonValue, C as CompletionModel, M as Message } from '../types-G12i56s2.js';
-import { A as Agent } from '../agent-x3YTyv5i.js';
-import { P as PromptResponse } from '../index-BxSbmOMr.js';
-import { E as EmbeddingModel } from '../types-BCTRUGex.js';
+import { A as Agent } from '../agent-DGYY_onS.js';
+import { l as AgentResponse } from '../index-DxB74ybJ.js';
+import { E as EmbeddingModel } from '../types-p55Jc-sr.js';
 import 'zod';
 import '../guardrails/index.js';
-import '../types-D9JpeAPF.js';
-import '../types-BrLbKnEz.js';
-import '../tool-Cohdv5GC.js';
-import '../middleware-Dp3d_pxi.js';
-import '../types-O-abTIin.js';
+import '../type-utils-CtHVDRn_.js';
+import '../types-BBEGKB9m.js';
+import '../middleware-KynLg_RW.js';
+import '../tool-C_8GudXn.js';
+import '../types-DulRCYdB.js';
+import '@modelcontextprotocol/sdk/client/sse.js';
+import '@modelcontextprotocol/sdk/client/stdio.js';
+import '@modelcontextprotocol/sdk/client/streamableHttp.js';
+import '../types-DhQfwiZn.js';
+import '../dynamic-tools-i9woysFa.js';
+import '../types-DT3nEemY.js';
+import '../retry-D3Ruy-ba.js';
 
 type EvalOutcome<Score = unknown> = {
     outcome: "pass";
@@ -381,11 +388,11 @@ declare function knowledgeRetention<Input, Output, Expected = unknown, const Nam
 }): EvalMetric<Input, Output, number, Expected, Name>;
 type ConversationSource = EvalTurn[] | Message[];
 
-type AgentEvalTargetOptions<Input, Output = PromptResponse, Expected = unknown> = {
+type AgentEvalTargetOptions<Input, Output = AgentResponse, Expected = unknown> = {
     prompt?: ((input: Input, testCase: EvalCase<Input, Expected>) => string | Message) | undefined;
-    output?: ((response: PromptResponse, testCase: EvalCase<Input, Expected>) => Output) | undefined;
+    output?: ((response: AgentResponse, testCase: EvalCase<Input, Expected>) => Output) | undefined;
 };
-declare function agentEvalTarget<Input, Expected = unknown>(agent: Agent, options?: AgentEvalTargetOptions<Input, PromptResponse, Expected>): EvalTarget<Input, PromptResponse, Expected>;
+declare function agentEvalTarget<Input, Expected = unknown>(agent: Agent, options?: AgentEvalTargetOptions<Input, AgentResponse, Expected>): EvalTarget<Input, AgentResponse, Expected>;
 declare function agentEvalTarget<Input, Output, Expected = unknown>(agent: Agent, options: AgentEvalTargetOptions<Input, Output, Expected>): EvalTarget<Input, Output, Expected>;
 
 type EvalOutputFormat = "pretty" | "json" | "quiet";
@@ -599,7 +606,7 @@ type DefinedEvalSuite<Cases extends readonly EvalCaseLike[], Output, Metrics ext
     target: EvalTarget<EvalCasesInput<Cases>, Output, EvalCasesExpected<Cases>>;
     metrics: Metrics;
 };
-type EvalSuiteTypeBuilder<Input, Output, Expected = unknown> = {
+type EvalMetricFactory<Input, Output, Expected = unknown> = {
     defineMetric<const Name extends string = string>(metric: EvalMetric<Input, Output, boolean, Expected, Name> & {
         dataType: "BOOLEAN";
     }): EvalMetric<Input, Output, boolean, Expected, Name>;
@@ -626,11 +633,11 @@ type EvalSuiteTypeBuilder<Input, Output, Expected = unknown> = {
     }): EvalMetric<Input, Output, Score, Expected, Name>;
 };
 declare function defineEvalCases<const Cases extends readonly EvalCaseLike[]>(cases: Cases): Cases;
-declare function defineEvalSuite<Input, Output, Expected = unknown>(): EvalSuiteTypeBuilder<Input, Output, Expected>;
+declare function defineEvalSuite<Input, Output, Expected = unknown>(): EvalMetricFactory<Input, Output, Expected>;
 declare function defineEvalSuite<const Cases extends readonly EvalCaseLike[], const Target extends (input: EvalCasesInput<Cases>, testCase: EvalCase<EvalCasesInput<Cases>, EvalCasesExpected<Cases>>) => unknown, const Metrics extends readonly EvalMetric<EvalCasesInput<Cases>, Awaited<ReturnType<Target>>, unknown, EvalCasesExpected<Cases>, string>[]>(options: DefinedEvalSuite<Cases, Awaited<ReturnType<Target>>, Metrics> & {
     target: Target;
 }): DefinedEvalSuite<Cases, Awaited<ReturnType<Target>>, Metrics> & {
     target: Target;
 };
 
-export { type AbstentionCategory, type AbstentionOptions, type AgentEvalTargetOptions, type AnswerRelevancyOptions, type AnyEvalMetric, type ContainsAllOptions, type ContainsAnyOptions, type ContainsListOptions, type ContainsOptions, type DefaultEvalActual, type DefinedEvalSuite, type DoesNotMatchOptions, EvalAssertionError, type EvalCase, type EvalCaseRequirements, type EvalCaseResult, type EvalCasesExpected, type EvalCasesForMetrics, type EvalCasesInput, type EvalCostCalculatorArgs, type EvalCostOptions, type EvalCostSummary, type EvalDataType, type EvalExpectations, type EvalExpectedOutcomes, type EvalExpectedTotals, type EvalMetadata, type EvalMetric, type EvalMetricArgs, type EvalMetricDescriptor, type EvalMetricResult, type EvalMetricResultFor, type EvalMetricScore, EvalOutcome, type EvalOutcomeStatus, type EvalOutputFormat, type EvalOutputWriters, type EvalReportArgs, type EvalReporter, type EvalRunContext, type EvalRunEndArgs, type EvalRunOptions, type EvalRunStartArgs, type EvalScoreDirection, type EvalScoreMap, type EvalScoreProjection, type EvalSuiteResult, type EvalSuiteTypeBuilder, type EvalTarget, type EvalTargetUsageSelector, type EvalTotals, type EvalTraceCarrier, type EvalTraceRef, type EvalTraceSelector, type EvalTraceSelectorArgs, type EvalTurn, type EvalUsageSummary, type ExactMatchOptions, type FaithfulnessOptions, type GEvalOptions, type GEvalParameter, type GEvalRubric, type HallucinationOptions, type JsonCorrectnessOptions, type KnowledgeRetentionOptions, type LlmJudgeOptions, type LlmScoreMetricScore, type LlmScoreOptions, type MatchesOptions, type MaxLengthOptions, type NotContainsOptions, type PrintEvalResultOptions, type PromptAlignmentOptions, type RequiredFieldsOptions, type RunEvalCliOptions, type RunEvalSuiteOptions, type SelectorOrValue, type SemanticSimilarityOptions, type SummarizationOptions, type TurnRelevancyOptions, type ValueSelector, abstention, agentEvalTarget, answerRelevancy, assertEvalOutcomes, assertEvalTotals, contains, containsAll, containsAny, defaultEvalTraceSelector, defineEvalCases, defineEvalSuite, defineMetric, doesNotMatch, evalExitCode, exactMatch, faithfulness, gEval, hallucination, jsonCorrectness, knowledgeRetention, llmJudge, llmScore, matches, maxLength, notContains, printEvalResult, projectEvalOutcome, promptAlignment, requiredFields, resolveEvalTraceRef, runEvalCli, runEvalSuite, selectPromptOutput, semanticSimilarity, summarization, turnRelevancy };
+export { type AbstentionCategory, type AbstentionOptions, type AgentEvalTargetOptions, type AnswerRelevancyOptions, type AnyEvalMetric, type ContainsAllOptions, type ContainsAnyOptions, type ContainsListOptions, type ContainsOptions, type DefaultEvalActual, type DefinedEvalSuite, type DoesNotMatchOptions, EvalAssertionError, type EvalCase, type EvalCaseRequirements, type EvalCaseResult, type EvalCasesExpected, type EvalCasesForMetrics, type EvalCasesInput, type EvalCostCalculatorArgs, type EvalCostOptions, type EvalCostSummary, type EvalDataType, type EvalExpectations, type EvalExpectedOutcomes, type EvalExpectedTotals, type EvalMetadata, type EvalMetric, type EvalMetricArgs, type EvalMetricDescriptor, type EvalMetricResult, type EvalMetricResultFor, type EvalMetricScore, EvalOutcome, type EvalOutcomeStatus, type EvalOutputFormat, type EvalOutputWriters, type EvalReportArgs, type EvalReporter, type EvalRunContext, type EvalRunEndArgs, type EvalRunOptions, type EvalRunStartArgs, type EvalScoreDirection, type EvalScoreMap, type EvalScoreProjection, type EvalSuiteResult, type EvalTarget, type EvalTargetUsageSelector, type EvalTotals, type EvalTraceCarrier, type EvalTraceRef, type EvalTraceSelector, type EvalTraceSelectorArgs, type EvalTurn, type EvalUsageSummary, type ExactMatchOptions, type FaithfulnessOptions, type GEvalOptions, type GEvalParameter, type GEvalRubric, type HallucinationOptions, type JsonCorrectnessOptions, type KnowledgeRetentionOptions, type LlmJudgeOptions, type LlmScoreMetricScore, type LlmScoreOptions, type MatchesOptions, type MaxLengthOptions, type NotContainsOptions, type PrintEvalResultOptions, type PromptAlignmentOptions, type RequiredFieldsOptions, type RunEvalCliOptions, type RunEvalSuiteOptions, type SelectorOrValue, type SemanticSimilarityOptions, type SummarizationOptions, type TurnRelevancyOptions, type ValueSelector, abstention, agentEvalTarget, answerRelevancy, assertEvalOutcomes, assertEvalTotals, contains, containsAll, containsAny, defaultEvalTraceSelector, defineEvalCases, defineEvalSuite, defineMetric, doesNotMatch, evalExitCode, exactMatch, faithfulness, gEval, hallucination, jsonCorrectness, knowledgeRetention, llmJudge, llmScore, matches, maxLength, notContains, printEvalResult, projectEvalOutcome, promptAlignment, requiredFields, resolveEvalTraceRef, runEvalCli, runEvalSuite, selectPromptOutput, semanticSimilarity, summarization, turnRelevancy };

package/dist/evals/index.js

@@ -1,30 +1,26 @@
 import {
-  ExtractorBuilder
-} from "../chunk-YLMDBQO4.js";
-import "../chunk-JI5YZWNT.js";
-import "../chunk-DALYP4PX.js";
-import "../chunk-BALBBVI3.js";
+  cancelAgentApproval
+} from "../chunk-IL3ZSNJI.js";
 import "../chunk-YK4WAAS4.js";
-import "../chunk-XUUY2L2D.js";
-import "../chunk-HYUYZEAJ.js";
-import "../chunk-MELYDKWQ.js";
-import "../chunk-2ODTMRHP.js";
-import "../chunk-UQX6EXWG.js";
-import "../chunk-MRQLEK2B.js";
+import "../chunk-6WGGTJE6.js";
+import "../chunk-R2LJSQUW.js";
+import "../chunk-5XKDW36Z.js";
 import {
   cosineSimilarity,
-  embedText
-} from "../chunk-MNWK2USR.js";
-import {
+  embedText,
   mapWithConcurrency
-} from "../chunk-OIMLU4SF.js";
-import "../chunk-Q25IWLBT.js";
-import "../chunk-MC3CCKEB.js";
+} from "../chunk-DZ3CKYTH.js";
+import {
+  Extractor
+} from "../chunk-J6GNQART.js";
+import "../chunk-I6G7BC42.js";
+import "../chunk-M3IOWB4Y.js";
+import "../chunk-RFDVKCIE.js";
 import {
   Usage
 } from "../chunk-ADH7NNCS.js";
-import "../chunk-WQKHFADH.js";
-import "../chunk-CWUJUSOS.js";
+import "../chunk-A3UBYTJT.js";
+import "../chunk-FNQB2JEH.js";
 
 // src/evals/advanced-metrics.ts
 import { z } from "zod";
@@ -58,7 +54,15 @@ function errorMessage(error) {
 
 // src/evals/judge.ts
 async function runJudge(args) {
-  const result = await new ExtractorBuilder(args.model, args.schema).instructions(args.instructions).temperature(0).retries(args.retries).build().extractWithUsage(args.prompt);
+  const extractor = new Extractor({
+    model: args.model,
+    outputSchema: args.schema,
+    instructions: args.instructions
+  });
+  const result = await extractor.extractResult(args.prompt, {
+    temperature: 0,
+    retries: args.retries <= 0 ? void 0 : { maxAttempts: Math.trunc(args.retries) + 1 }
+  });
   return { data: result.data, usage: result.usage };
 }
 function addUsage(...values) {
@@ -1119,7 +1123,11 @@ function contentText(content) {
 function agentEvalTarget(agent, options = {}) {
   return async (input, testCase) => {
     const prompt = options.prompt?.(input, testCase) ?? String(input);
-    const response = await agent.prompt(prompt).send();
+    const response = await agent.generate(prompt);
+    if (response.status === "approval_required") {
+      await cancelAgentApproval(response, "Agent eval targets cannot suspend for tool approval.");
+      throw new Error("Agent eval targets cannot suspend for tool approval.");
+    }
     return options.output === void 0 ? response : options.output(response, testCase);
   };
 }
@@ -2006,16 +2014,21 @@ function semanticSimilarity(options) {
   };
 }
 function llmJudge(options) {
-  const extractor = new ExtractorBuilder(options.model, options.schema).instructions(
-    options.instructions ?? "Judge the eval case by the requested schema. Submit the judgment using the schema."
-  ).retries(options.retries ?? 0).build();
+  const extractor = new Extractor({
+    model: options.model,
+    outputSchema: options.schema,
+    instructions: options.instructions ?? "Judge the eval case by the requested schema. Submit the judgment using the schema."
+  });
   return {
     name: options.name ?? "llm_judge",
     required: options.required ?? true,
     async evaluate(args) {
       try {
-        const result = await extractor.extractWithUsage(
-          await resolveJudgePrompt(options.prompt, args)
+        const result = await extractor.extractResult(
+          await resolveJudgePrompt(options.prompt, args),
+          {
+            retries: evalExtractionRetries(options.retries)
+          }
         );
         return options.passes(result.data) ? EvalOutcome.pass(result.data, { usage: result.usage }) : EvalOutcome.fail(result.data, { usage: result.usage });
       } catch (error) {
@@ -2026,18 +2039,17 @@ function llmJudge(options) {
 }
 function llmScore(options) {
   const criteria = Array.isArray(options.criteria) ? options.criteria.join("\n") : options.criteria;
-  const extractor = new ExtractorBuilder(
-    options.model,
-    z2.object({
+  const extractor = new Extractor({
+    model: options.model,
+    outputSchema: z2.object({
       score: z2.number(),
       feedback: z2.string()
-    })
-  ).instructions(
-    options.instructions ?? `Score the eval case against these criteria:
+    }),
+    instructions: options.instructions ?? `Score the eval case against these criteria:
 ${criteria}
 
 Return a score between 0 and 1 and brief feedback.`
-  ).retries(options.retries ?? 0).build();
+  });
   return {
     name: options.name ?? "llm_score",
     required: options.required ?? true,
@@ -2047,8 +2059,11 @@ Return a score between 0 and 1 and brief feedback.`
     threshold: options.threshold,
     async evaluate(args) {
       try {
-        const result = await extractor.extractWithUsage(
-          await resolveJudgePrompt(options.prompt, args)
+        const result = await extractor.extractResult(
+          await resolveJudgePrompt(options.prompt, args),
+          {
+            retries: evalExtractionRetries(options.retries)
+          }
         );
         const score = result.data;
         if (score.score < 0 || score.score > 1) {
@@ -2064,6 +2079,13 @@ Return a score between 0 and 1 and brief feedback.`
     }
   };
 }
+function evalExtractionRetries(retries) {
+  const retryCount = Math.max(0, Math.trunc(retries ?? 0));
+  if (retryCount === 0) {
+    return void 0;
+  }
+  return { maxAttempts: retryCount + 1 };
+}
 
 // src/evals/suite.ts
 function defineEvalCases(cases) {