@wix/evalforge-evaluator 0.140.0 → 0.142.0

package/build/index.js

@@ -287,13 +287,16 @@ function resolveSystemAssertion(assertionId, params) {
         }
       };
       break;
-    case "build_passed":
+    case "build_passed": {
+      const rawCmd = params?.command;
+      const command = typeof rawCmd === "string" && (0, import_evalforge_types.isAllowedBuildCommandString)(rawCmd) ? rawCmd.trim() : void 0;
       baseAssertion = {
         type: "build_passed",
-        command: params?.command ?? void 0,
+        ...command !== void 0 && { command },
         expectedExitCode: params?.expectedExitCode ?? void 0
       };
       break;
+    }
     case "time_limit":
       baseAssertion = {
         type: "time_limit",
@@ -4754,6 +4757,7 @@ var import_crypto5 = require("crypto");
 async function runEvaluationLoop(scenarioItems, evalData, callbacks) {
   const runsPerScenario = evalData.evalRun.runsPerScenario ?? 1;
   let completedExecutions = 0;
+  let erroredExecutions = 0;
   const totalExecutions = scenarioItems.length * runsPerScenario;
   const targetId = evalData.evalRun.presetId ?? evalData.agent?.id ?? evalData.evalRun.id;
   const targetName = evalData.presetName ?? evalData.agent?.name ?? "";
@@ -4790,6 +4794,7 @@ async function runEvaluationLoop(scenarioItems, evalData, callbacks) {
           iterationIndex
         };
         await callbacks.addResult(errorResult);
+        erroredExecutions++;
       }
       if (scenarioResult !== null) {
         await callbacks.addResult({ ...scenarioResult, iterationIndex });
@@ -4797,7 +4802,7 @@ async function runEvaluationLoop(scenarioItems, evalData, callbacks) {
       completedExecutions++;
     }
   }
-  return { completedExecutions, totalExecutions };
+  return { completedExecutions, totalExecutions, erroredExecutions };
 }
 
 // src/error-reporter.ts
@@ -4981,50 +4986,46 @@ async function runEvaluation(projectId2, evalRunId2) {
     );
   }
   const skillNames = evalData.skills.map((s) => s.name).join(", ");
-  const { completedExecutions, totalExecutions } = await runEvaluationLoop(
-    scenarioItems,
-    evalData,
-    {
-      runScenario: (scenario, template, resolvedAssertions) => {
-        state.currentPhase = ExecutionPhase.EXECUTE_SKILL;
-        state.currentContext = {
-          projectId: projectId2,
-          evalRunId: evalRunId2,
-          scenarioId: scenario.id,
-          scenarioName: scenario.name,
-          presetId: evalData.evalRun.presetId,
-          presetName: evalData.presetName,
-          agentId: agent?.id,
-          agentName: agent?.name
-        };
-        return runScenario(
-          config,
-          evalRunId2,
-          scenario,
-          evalData,
-          template,
-          resolvedAssertions
-        );
-      },
-      addResult: async (result) => {
-        state.currentPhase = ExecutionPhase.ADD_RESULT;
-        state.currentContext = {
-          ...state.currentContext,
-          resultId: result.id
-        };
-        await api.addResult(projectId2, evalRunId2, result);
-      },
-      onProgress: (completed, total, iterLabel) => {
-        console.log(
-          "[Evaluator] Running scenario with preset:",
-          evalData.presetName,
-          skillNames ? `(${skillNames})` : "",
-          agent ? `with agent: ${agent.name}` : "",
-          `(${completed}/${total})${iterLabel}`
-        );
-      }
+  const { completedExecutions, totalExecutions, erroredExecutions } = await runEvaluationLoop(scenarioItems, evalData, {
+    runScenario: (scenario, template, resolvedAssertions) => {
+      state.currentPhase = ExecutionPhase.EXECUTE_SKILL;
+      state.currentContext = {
+        projectId: projectId2,
+        evalRunId: evalRunId2,
+        scenarioId: scenario.id,
+        scenarioName: scenario.name,
+        presetId: evalData.evalRun.presetId,
+        presetName: evalData.presetName,
+        agentId: agent?.id,
+        agentName: agent?.name
+      };
+      return runScenario(
+        config,
+        evalRunId2,
+        scenario,
+        evalData,
+        template,
+        resolvedAssertions
+      );
+    },
+    addResult: async (result) => {
+      state.currentPhase = ExecutionPhase.ADD_RESULT;
+      state.currentContext = {
+        ...state.currentContext,
+        resultId: result.id
+      };
+      await api.addResult(projectId2, evalRunId2, result);
+    },
+    onProgress: (completed, total, iterLabel) => {
+      console.log(
+        "[Evaluator] Running scenario with preset:",
+        evalData.presetName,
+        skillNames ? `(${skillNames})` : "",
+        agent ? `with agent: ${agent.name}` : "",
+        `(${completed}/${total})${iterLabel}`
+      );
     }
-  );
+  });
   state.currentPhase = ExecutionPhase.UPDATE_STATUS;
   state.currentContext = {
     projectId: projectId2,
@@ -5032,14 +5033,16 @@ async function runEvaluation(projectId2, evalRunId2) {
     completedExecutions,
     totalExecutions
   };
+  const allFailed = totalExecutions > 0 && erroredExecutions === totalExecutions;
+  const finalStatus = allFailed ? import_evalforge_types15.EvalStatus.FAILED : import_evalforge_types15.EvalStatus.COMPLETED;
   try {
     await api.updateEvalRun(projectId2, evalRunId2, {
-      status: import_evalforge_types15.EvalStatus.COMPLETED,
+      status: finalStatus,
       completedAt: (/* @__PURE__ */ new Date()).toISOString()
     });
   } catch (updateErr) {
     throw new Error(
-      `[${ExecutionPhase.UPDATE_STATUS}] Failed to update eval run status to COMPLETED: ${updateErr instanceof Error ? updateErr.message : String(updateErr)}`
+      `[${ExecutionPhase.UPDATE_STATUS}] Failed to update eval run status to ${finalStatus}: ${updateErr instanceof Error ? updateErr.message : String(updateErr)}`
     );
   }
 }