@alis-build/alis-evals-v1 1.24.0 → 1.27.0

package/alis/evals/v1/evaluation_pb.js

@@ -1,7 +1,31 @@
-// @generated by protoc-gen-es v2.13.0 with parameter "target=js+dts,import_extension=js,json_types=true"
+// @generated by protoc-gen-es v2.16.0 with parameter "target=js+dts,import_extension=js,json_types=true"
 // @generated from file alis/evals/v1/evaluation.proto (package alis.evals.v1, syntax proto3)
 /* eslint-disable */
 
+// Runs post-deploy tests and evaluations and records the outcome of each
+// finished suite as a `Run` (`runs/{run_id}`).
+//
+// `TestService` has one RPC per kind of run. Each starts a
+// `google.longrunning.Operation` whose response lists the names of the runs
+// it produced:
+//
+// - `RunIntegrationTest`: integration cases scored by deterministic checks.
+// - `RunLoadTest`: generated traffic at a `RunLoadTestRequest.Mode` preset,
+//   with Service Level Objective (SLO) checks on latency, errors, and
+//   throughput.
+// - `RunAgentEval`: agent evaluation cases scored by deterministic checks,
+//   LLM-as-judge scores, rubric dimensions, and ADK prebuilt metrics.
+// - `RunInfraObservation`: Cloud Monitoring snapshots of declared Cloud Run
+//   and Spanner targets, without generating load.
+//
+// Requests select cases by id: `suite.case` selects one case, `suite` selects
+// every case in that suite, and an empty list runs every registered case. An
+// operation completes successfully once execution finishes, even when cases
+// fail, so read `Run.status` for the outcome. Runs, cases, checks, and
+// metrics all report a `Status`. Each completed run is published as a
+// `RunPublishedEvent` for persistence and downstream consumers such as
+// BigQuery or Pub/Sub.
+
 import { enumDesc, fileDesc, messageDesc, serviceDesc, tsEnum } from "@bufbuild/protobuf/codegenv2";
 import { file_google_longrunning_operations } from "@alis-build/google-longrunning/google/longrunning/operations_pb.js";
 import { file_google_protobuf_duration, file_google_protobuf_struct, file_google_protobuf_timestamp } from "@bufbuild/protobuf/wkt";
@@ -27,6 +51,8 @@ export const Run_TypeSchema = /*@__PURE__*/
   enumDesc(file_alis_evals_v1_evaluation, 0, 0);
 
 /**
+ * Identifies which kind of test or evaluation produced a Run.
+ *
  * @generated from enum alis.evals.v1.Run.Type
  */
 export const Run_Type = /*@__PURE__*/
@@ -249,6 +275,10 @@ export const RunLoadTestRequest_ModeSchema = /*@__PURE__*/
   enumDesc(file_alis_evals_v1_evaluation, 16, 0);
 
 /**
+ * Load intensity presets, ordered as an ascending scale from minimal to
+ * ludicrous. Higher modes apply more concurrent load; the exact parameters
+ * are environment-specific.
+ *
  * @generated from enum alis.evals.v1.RunLoadTestRequest.Mode
  */
 export const RunLoadTestRequest_Mode = /*@__PURE__*/
@@ -331,6 +361,12 @@ export const InfraTargetRoleSchema = /*@__PURE__*/
   enumDesc(file_alis_evals_v1_evaluation, 0);
 
 /**
+ * Role of an infrastructure target within a load or observation case.
+ *
+ * ENTRY marks the service under direct test (for example the Cloud Run service
+ * receiving generated load). DEPENDENCY marks downstream services or data stores
+ * observed for correlation (for example internal Cloud Run hops or Spanner).
+ *
  * @generated from enum alis.evals.v1.InfraTargetRole
  */
 export const InfraTargetRole = /*@__PURE__*/
@@ -343,6 +379,12 @@ export const InfraFetchStatusSchema = /*@__PURE__*/
   enumDesc(file_alis_evals_v1_evaluation, 1);
 
 /**
+ * Outcome of fetching infrastructure metrics for one target from Cloud
+ * Monitoring.
+ *
+ * Partial per-metric failure may still yield OK when at least one metric
+ * succeeded; details are recorded in fetch_message on the snapshot.
+ *
  * @generated from enum alis.evals.v1.InfraFetchStatus
  */
 export const InfraFetchStatus = /*@__PURE__*/
@@ -355,6 +397,8 @@ export const InfraKindSchema = /*@__PURE__*/
   enumDesc(file_alis_evals_v1_evaluation, 2);
 
 /**
+ * Infrastructure resource kind referenced by infra SLO checks and snapshots.
+ *
  * @generated from enum alis.evals.v1.InfraKind
  */
 export const InfraKind = /*@__PURE__*/
@@ -367,12 +411,21 @@ export const StatusSchema = /*@__PURE__*/
   enumDesc(file_alis_evals_v1_evaluation, 3);
 
 /**
+ * Outcome status for runs, cases, checks, metrics, and rubric dimensions.
+ *
  * @generated from enum alis.evals.v1.Status
  */
 export const Status = /*@__PURE__*/
   tsEnum(StatusSchema);
 
 /**
+ * Executes post-deploy tests and evaluations. Each RPC starts a long-running
+ * operation that runs bundled cases for one kind of test. One operation may
+ * produce multiple Run resources (one per completed suite). When the operation
+ * completes, the response returns the resource names of all resulting Runs
+ * (`runs/{run_id}`). Each full Run payload is published for persistence and
+ * event consumers (for example BigQuery or Pub/Sub).
+ *
  * @generated from service alis.evals.v1.TestService
  */
 export const TestService = /*@__PURE__*/

package/package.json

@@ -26,5 +26,5 @@
   },
   "type": "module",
   "types": "./index.d.ts",
-  "version": "1.24.0"
+  "version": "1.27.0"
 }