agentv 5.2.0-next.1 → 5.3.0-next.1

package/dist/{chunk-46K2OET3.js → chunk-BHKQHG26.js}

@@ -6,7 +6,7 @@ import {
   toSnakeCaseDeep,
   writeArtifactsFromResults,
   writePerTestArtifacts
-} from "./chunk-3S6A2RKR.js";
+} from "./chunk-6ZZCDZPD.js";
 
 // src/commands/eval/artifact-writer.ts
 import path2 from "node:path";
@@ -1020,4 +1020,4 @@ export {
   writePerTestArtifacts2 as writePerTestArtifacts,
   writeArtifactsFromResults2 as writeArtifactsFromResults
 };
-//# sourceMappingURL=chunk-46K2OET3.js.map
+//# sourceMappingURL=chunk-BHKQHG26.js.map

package/dist/{chunk-T4AD3H3Y.js → chunk-FVR4RQFK.js}

@@ -1,7 +1,7 @@
 import { createRequire } from 'node:module'; const require = createRequire(import.meta.url);
 import {
   writeArtifactsFromResults
-} from "./chunk-46K2OET3.js";
+} from "./chunk-BHKQHG26.js";
 import {
   RunBudgetTracker,
   buildWipBranchName,
@@ -18,7 +18,7 @@ import {
   resolveResultsRepoRunsDir,
   setupWipWorktree,
   syncResultsRepoForProject
-} from "./chunk-ALVZQUZP.js";
+} from "./chunk-V52ATPTT.js";
 import {
   AGENTV_LOCAL_CONFIG_FILE_NAME,
   AGENTV_LOCAL_CONFIG_YML_FILE_NAME,
@@ -59,6 +59,7 @@ import {
   loadTestSuite,
   loadTestSuiteFromYamlObject,
   normalizeComposableConfigGraph,
+  normalizeMetricsArtifactWire,
   normalizeResultRow,
   normalizeTargetDefinition,
   parseJsonlResults,
@@ -83,7 +84,7 @@ import {
   traceEnvelopeToTraceSummary,
   traceEnvelopeToTranscriptMessages,
   writeInitialRunSummaryArtifact
-} from "./chunk-3S6A2RKR.js";
+} from "./chunk-6ZZCDZPD.js";
 
 // src/commands/eval/shared.ts
 import { constants } from "node:fs";
@@ -346,7 +347,7 @@ async function findRepoRoot(start) {
 // package.json
 var package_default = {
   name: "agentv",
-  version: "5.2.0-next.1",
+  version: "5.3.0-next.1",
   description: "CLI entry point for AgentV",
   type: "module",
   repository: {
@@ -1053,7 +1054,9 @@ function hydrateTrace(baseDir, record, options) {
 }
 function hydrateManifestRecord(baseDir, record, options) {
   const grading = readOptionalJson(baseDir, record.grading_path);
-  const metrics = readOptionalJson(baseDir, record.metrics_path);
+  const metrics = normalizeMetricsArtifactWire(
+    readOptionalJson(baseDir, record.metrics_path)
+  );
   const timing = metrics ?? readOptionalJson(baseDir, record.timing_path);
   const testId = record.test_id ?? "unknown";
   const gradingAssertions = grading ? collectComponentAssertions(grading.component_results) : void 0;
@@ -2853,18 +2856,63 @@ var UNSUPPORTED_PROMPTFOO_ASSERTION_TYPES = /* @__PURE__ */ new Set([
   "human",
   "max-score",
   "tool-call-f1",
-  "skill-used",
-  "trajectory:goal-success",
-  "trajectory:tool-args-match",
-  "trajectory:step-count",
-  "trajectory:tool-sequence",
-  "trajectory:tool-used",
   "trace-error-spans",
   "trace-span-count",
   "trace-span-duration",
   "search-rubric",
   "word-count"
 ]);
+function formatJsonValue(value) {
+  if (typeof value === "string") return value;
+  try {
+    return JSON.stringify(value);
+  } catch {
+    return String(value);
+  }
+}
+function hasToolTrajectoryLatencyCheck(value) {
+  const expected = value.expected;
+  return Array.isArray(expected) && expected.some(
+    (item) => isObject(item) && (item.max_duration_ms !== void 0 || item.maxDurationMs !== void 0)
+  );
+}
+function staleSkillTriggerMessage(value) {
+  const skill = typeof value.skill === "string" && value.skill.trim() ? value.skill.trim() : "";
+  const shouldTrigger = value.should_trigger !== false;
+  if (!skill) {
+    return "Authored assertion type 'skill-trigger' has been removed. Use 'skill-used' with value: <skill> for expected skill use, or 'not-skill-used' with value: <skill> when the skill must not be used.";
+  }
+  const replacementType = shouldTrigger ? "skill-used" : "not-skill-used";
+  return `Authored assertion type 'skill-trigger' has been removed. Replace skill: ${skill} with type: ${replacementType}, value: ${skill}.`;
+}
+function staleToolTrajectoryMessage(value) {
+  const mode = typeof value.mode === "string" ? value.mode : void 0;
+  if (hasToolTrajectoryLatencyCheck(value)) {
+    return "Authored assertion type 'tool-trajectory' has been removed. Per-tool latency checks such as max_duration_ms have no Promptfoo trajectory:* equivalent in AgentV yet; use a custom script assertion or track this as unsupported future scope.";
+  }
+  if (mode === "any_order" && isObject(value.minimums)) {
+    const [toolName, min] = Object.entries(value.minimums)[0] ?? [];
+    const example = typeof toolName === "string" ? ` For example: type: trajectory:tool-used, value: { name: ${toolName}, min: ${formatJsonValue(min)} }.` : "";
+    return `Authored assertion type 'tool-trajectory' has been removed. Replace mode: any_order minimums with one Promptfoo trajectory:tool-used assertion per tool.${example}`;
+  }
+  if ((mode === "in_order" || mode === "exact") && Array.isArray(value.expected)) {
+    const steps = value.expected.filter(isObject).map((item) => item.tool).filter((tool) => typeof tool === "string");
+    const hasArgs = value.expected.some(
+      (item) => isObject(item) && item.args !== void 0 && item.args !== "any"
+    );
+    const sequenceHint = steps.length > 0 ? ` Use type: trajectory:tool-sequence, value: { mode: ${mode}, steps: ${formatJsonValue(steps)} }.` : " Use type: trajectory:tool-sequence with value: { mode, steps }.";
+    const argsHint = hasArgs ? " Move expected args checks to trajectory:tool-args-match with Promptfoo mode: partial or exact." : "";
+    return `Authored assertion type 'tool-trajectory' has been removed.${sequenceHint}${argsHint}`;
+  }
+  return "Authored assertion type 'tool-trajectory' has been removed. Use Promptfoo trajectory:* assertions: trajectory:tool-used for tool presence/counts, trajectory:tool-sequence for in_order/exact steps, and trajectory:tool-args-match for argument checks. AgentV-specific latency checks are unsupported future scope.";
+}
+function staleAuthoredAssertionMessage(value) {
+  if (typeof value.type !== "string") return void 0;
+  const type = value.type.replace(/_/g, "-");
+  if (type === "skill-trigger") return staleSkillTriggerMessage(value);
+  if (type === "tool-trajectory") return staleToolTrajectoryMessage(value);
+  return void 0;
+}
 var VALID_TEST_FILE_EXTENSIONS = /* @__PURE__ */ new Set([
   ".csv",
   ".cjs",
@@ -4611,6 +4659,16 @@ function validateAssertArray(assertField, location, filePath, errors, customAsse
       continue;
     }
     const typeValue = rawTypeValue.replace(/_/g, "-");
+    const staleMessage = staleAuthoredAssertionMessage(item);
+    if (staleMessage) {
+      errors.push({
+        severity: "error",
+        filePath,
+        location: `${itemLocation}.type`,
+        message: staleMessage
+      });
+      continue;
+    }
     const replacement = REMOVED_ASSERTION_TYPE_REPLACEMENTS.get(typeValue);
     if (replacement) {
       errors.push({
@@ -4631,7 +4689,7 @@ function validateAssertArray(assertField, location, filePath, errors, customAsse
       });
       continue;
     }
-    if (!isGraderKind(typeValue) && !PROMPTFOO_ASSERTION_TYPES.has(typeValue) && !customAssertionTypes.has(typeValue)) {
+    if (!isGraderKind(typeValue) && !isGraderKind(baseTypeValue) && !PROMPTFOO_ASSERTION_TYPES.has(typeValue) && !customAssertionTypes.has(typeValue)) {
       errors.push({
         severity: "warning",
         filePath,
@@ -7527,7 +7585,7 @@ async function runEvalCommand(input) {
   const resolvedExperiment = resolveExperimentForRun(options.experiment);
   const evalPathInputs = input.testFiles.length > 0 ? [...input.testFiles] : [];
   if (evalPathInputs.length === 0 && process.stdin.isTTY) {
-    const { launchInteractiveWizard } = await import("./interactive-36ZNMQB2.js");
+    const { launchInteractiveWizard } = await import("./interactive-RUY3OCBI.js");
     await launchInteractiveWizard();
     return void 0;
   }
@@ -7867,7 +7925,7 @@ async function runEvalCommand(input) {
   });
   let transcriptProviderFactory;
   if (options.transcript) {
-    const { TranscriptProvider } = await import("./dist-DEPJOMCA.js");
+    const { TranscriptProvider } = await import("./dist-6Z7U473R.js");
     const transcriptProvider = await TranscriptProvider.fromFile(options.transcript);
     const totalTests = [...fileMetadata.values()].reduce(
       (sum, meta) => sum + meta.testCases.length,
@@ -8133,7 +8191,7 @@ ${formatRequiredVersionFailureNote(requiredVersionCheck)}`);
       const sourceTests = activeSourceTests;
       const taskBundleTargets = buildTaskBundleTargetSelections(activeTestFiles, fileMetadata);
       if (isResumeAppend) {
-        const { writePerTestArtifacts } = await import("./artifact-writer-SN6UNL5G.js");
+        const { writePerTestArtifacts } = await import("./artifact-writer-JFNIPMKW.js");
         await writePerTestArtifacts(allResults, runDir, {
           experiment: normalizeExperimentName(options.experiment),
           runId: path12.basename(runDir),
@@ -8351,4 +8409,4 @@ export {
   filterByCategory,
   runEvalCommand
 };
-//# sourceMappingURL=chunk-T4AD3H3Y.js.map
+//# sourceMappingURL=chunk-FVR4RQFK.js.map