anpord 0.1.21 → 0.1.22

package/dist/{compiler-Cy85vNbp.cjs → compiler-B1UrHfCL.cjs}

@@ -1,6 +1,6 @@
 const require_eval_judges = require("./eval-judges-1zF4PVqV.cjs");
 const require_api_mocks = require("./api-mocks-CgiC7T9U.cjs");
-const require_evals_api = require("./evals-api-CNoLlXL5.cjs");
+const require_evals_api = require("./evals-api-DUiZWsyC.cjs");
 const require_source = require("./source-aW5nSo2h.cjs");
 let effect = require("effect");
 let node_fs_promises = require("node:fs/promises");
@@ -437,6 +437,16 @@ function suite(definition) {
 		value: file
 	});
 }
+function evalCase(definition) {
+	const { prompt, tasks, trials, ...subject } = definition;
+	return suite({
+		cases: [subject],
+		name: subject.name,
+		prompt,
+		tasks,
+		trials
+	});
+}
 const sourceUrlOf = (definition) => definition[SOURCE_URL];
 //#endregion
 //#region src/evals/locate.ts
@@ -813,6 +823,7 @@ const compileRefEffect = (ref) => effect.Effect.gen(function* () {
 			name: subject.name,
 			prepare,
 			...sourceFor(definition, subject),
+			...subject.tags === void 0 ? {} : { tags: subject.tags },
 			user: subject.user ?? null,
 			validator,
 			variables: subject.variables ?? {},
@@ -855,6 +866,12 @@ Object.defineProperty(exports, "compileEvalEffect", {
 		return compileEvalEffect;
 	}
 });
+Object.defineProperty(exports, "evalCase", {
+	enumerable: true,
+	get: function() {
+		return evalCase;
+	}
+});
 Object.defineProperty(exports, "sourceUrlOf", {
 	enumerable: true,
 	get: function() {

package/dist/{compiler-DSQRHvXS.mjs → compiler-D9jidpZ0.mjs}

@@ -1,6 +1,6 @@
 import { t as EvalJudge } from "./eval-judges-Bee_ABJt.mjs";
 import { r as API_PROGRAM } from "./api-mocks-BYWvJp9v.mjs";
-import { c as HarnessProfile, d as profileFitsHarness, f as EvalSourceFiles, l as PROFILE_LIMITS, n as PublicStartEvalRequest, o as EvalValidator, s as EnvName, u as ProfilePath } from "./evals-api-CK3F1KoI.mjs";
+import { c as HarnessProfile, d as profileFitsHarness, f as EvalSourceFiles, l as PROFILE_LIMITS, n as PublicStartEvalRequest, o as EvalValidator, s as EnvName, u as ProfilePath } from "./evals-api-NLUUV9tM.mjs";
 import { r as repo } from "./source-BP11pzVV.mjs";
 import { Data, Effect, Option, Schema } from "effect";
 import { mkdtemp, readFile, readdir, rm, stat, writeFile } from "node:fs/promises";
@@ -437,6 +437,16 @@ function suite(definition) {
 		value: file
 	});
 }
+function evalCase(definition) {
+	const { prompt, tasks, trials, ...subject } = definition;
+	return suite({
+		cases: [subject],
+		name: subject.name,
+		prompt,
+		tasks,
+		trials
+	});
+}
 const sourceUrlOf = (definition) => definition[SOURCE_URL];
 //#endregion
 //#region src/evals/locate.ts
@@ -813,6 +823,7 @@ const compileRefEffect = (ref) => Effect.gen(function* () {
 			name: subject.name,
 			prepare,
 			...sourceFor(definition, subject),
+			...subject.tags === void 0 ? {} : { tags: subject.tags },
 			user: subject.user ?? null,
 			validator,
 			variables: subject.variables ?? {},
@@ -837,4 +848,4 @@ const refOfPath = (path) => ({
 const compileEvalEffect = (path) => compileRefEffect(refOfPath(path));
 const compileEval = (path) => Effect.runPromise(compileEvalEffect(path));
 //#endregion
-export { suite as a, sourceUrlOf as i, compileEval as n, compileEvalEffect as r, compileDefinition as t };
+export { sourceUrlOf as a, evalCase as i, compileEval as n, suite as o, compileEvalEffect as r, compileDefinition as t };

package/dist/config.cjs

@@ -1,5 +1,5 @@
 Object.defineProperty(exports, Symbol.toStringTag, { value: "Module" });
-const require_client = require("./client-BZDftC1x.cjs");
+const require_client = require("./client-CWVAt_wk.cjs");
 let effect = require("effect");
 //#region src/client/config.ts
 const apiKeyConfig = effect.Config.redacted("ANPORD_API_KEY");

package/dist/config.d.cts

@@ -1,4 +1,4 @@
-import { t as AnpordApi } from "./client-CsGxLUZq.cjs";
+import { t as AnpordApi } from "./client-CE9VurDH.cjs";
 import { Config, Layer } from "effect";
 //#region src/client/config.d.ts
 declare const apiKeyConfig: Config.Config<import("effect/Redacted").Redacted<string>>;

package/dist/config.d.mts

@@ -1,4 +1,4 @@
-import { t as AnpordApi } from "./client-C7iEhiDm.mjs";
+import { t as AnpordApi } from "./client-EdYxRHX0.mjs";
 import { Config, Layer } from "effect";
 //#region src/client/config.d.ts
 declare const apiKeyConfig: Config.Config<import("effect/Redacted").Redacted<string>>;

package/dist/config.mjs

@@ -1,4 +1,4 @@
-import { n as DEFAULT_BASE_URL, r as layer } from "./client-D0l3kwWV.mjs";
+import { n as DEFAULT_BASE_URL, r as layer } from "./client-CfPf0hIc.mjs";
 import { Config, Effect, Layer } from "effect";
 //#region src/client/config.ts
 const apiKeyConfig = Config.redacted("ANPORD_API_KEY");

package/dist/{eval-turns-CTCoSuB6.d.cts → eval-turns-Uwe0G7PS.d.cts}

@@ -17,7 +17,7 @@ declare const EvalJudge: Schema.Union<[Schema.Struct<{
   }>;
 }>, Schema.Struct<{
   provider: Schema.optional<typeof Schema.Never>;
-  harness: Schema.SchemaClass<"cursor" | "codex" | "opencode" | "pi" | "fx" | "claude" | "gemini" | "qwen", "cursor" | "codex" | "opencode" | "pi" | "fx" | "claude" | "gemini" | "qwen", never>;
+  harness: Schema.SchemaClass<"codex" | "opencode" | "pi" | "fx" | "claude" | "gemini" | "qwen" | "cursor", "codex" | "opencode" | "pi" | "fx" | "claude" | "gemini" | "qwen" | "cursor", never>;
   kind: Schema.Literal<["judge"]>;
   name: Schema.filter<Schema.filter<typeof Schema.String>>;
   model: Schema.filter<Schema.filter<typeof Schema.String>>;

package/dist/{eval-turns-CTCoSuB6.d.mts → eval-turns-Uwe0G7PS.d.mts}

@@ -17,7 +17,7 @@ declare const EvalJudge: Schema.Union<[Schema.Struct<{
   }>;
 }>, Schema.Struct<{
   provider: Schema.optional<typeof Schema.Never>;
-  harness: Schema.SchemaClass<"cursor" | "codex" | "opencode" | "pi" | "fx" | "claude" | "gemini" | "qwen", "cursor" | "codex" | "opencode" | "pi" | "fx" | "claude" | "gemini" | "qwen", never>;
+  harness: Schema.SchemaClass<"codex" | "opencode" | "pi" | "fx" | "claude" | "gemini" | "qwen" | "cursor", "codex" | "opencode" | "pi" | "fx" | "claude" | "gemini" | "qwen" | "cursor", never>;
   kind: Schema.Literal<["judge"]>;
   name: Schema.filter<Schema.filter<typeof Schema.String>>;
   model: Schema.filter<Schema.filter<typeof Schema.String>>;

package/dist/eval.cjs

@@ -1,4 +1,4 @@
 Object.defineProperty(exports, Symbol.toStringTag, { value: "Module" });
-const require_compiler = require("./compiler-Cy85vNbp.cjs");
+const require_compiler = require("./compiler-B1UrHfCL.cjs");
 exports.compileDefinition = require_compiler.compileDefinition;
 exports.compileEval = require_compiler.compileEval;

package/dist/eval.d.cts

@@ -1,4 +1,4 @@
-import { h as PublicStartEvalRequest, i as EvalDefinition } from "./types-Dvd_z5ge.cjs";
+import { g as PublicStartEvalRequest, i as EvalDefinition } from "./types-C1Np-yYM.cjs";
 import { Effect } from "effect";
 //#region src/evals/compiler.d.ts
 declare const compileDefinition: (definition: EvalDefinition) => Promise<PublicStartEvalRequest>;

package/dist/eval.d.mts

@@ -1,4 +1,4 @@
-import { h as PublicStartEvalRequest, i as EvalDefinition } from "./types-D-h67MVJ.mjs";
+import { g as PublicStartEvalRequest, i as EvalDefinition } from "./types-B1rRGWeL.mjs";
 import { Effect } from "effect";
 //#region src/evals/compiler.d.ts
 declare const compileDefinition: (definition: EvalDefinition) => Promise<PublicStartEvalRequest>;

package/dist/eval.mjs

@@ -1,2 +1,2 @@
-import { n as compileEval, t as compileDefinition } from "./compiler-DSQRHvXS.mjs";
+import { n as compileEval, t as compileDefinition } from "./compiler-D9jidpZ0.mjs";
 export { compileDefinition, compileEval };

package/dist/{evals-Dw2Ek-mx.d.cts → evals-M2G4zvnS.d.cts}

@@ -91,14 +91,9 @@ declare const EvalTrial: Schema.Struct<{
     sha256: Schema.filter<typeof Schema.String>;
   }>>>;
   validations: Schema.optional<Schema.refine<readonly {
-    readonly name: string;
-    readonly id: string;
-    readonly kind: "command" | "judge" | "code";
-    readonly status: "passed" | "running" | "failed" | "error" | "queued" | "skipped";
+    readonly kind: "judge" | "command" | "code";
     readonly message: string;
-    readonly startedAt: number | null;
-    readonly index: number;
-    readonly exitCode: number | null;
+    readonly name: string;
     readonly output: {
       readonly text: string;
       readonly truncated: boolean;
@@ -111,6 +106,11 @@ declare const EvalTrial: Schema.Struct<{
       readonly format: "text" | "json";
       readonly state: "captured" | "disabled" | "unavailable";
     };
+    readonly id: string;
+    readonly status: "passed" | "running" | "failed" | "error" | "queued" | "skipped";
+    readonly startedAt: number | null;
+    readonly index: number;
+    readonly exitCode: number | null;
     readonly error: {
       readonly text: string;
       readonly truncated: boolean;
@@ -137,8 +137,6 @@ declare const EvalTrial: Schema.Struct<{
       readonly state: "captured" | "disabled" | "unavailable";
     } | undefined;
     readonly calls: readonly {
-      readonly startedAt: number;
-      readonly index: number;
       readonly output: {
         readonly text: string;
         readonly truncated: boolean;
@@ -151,6 +149,8 @@ declare const EvalTrial: Schema.Struct<{
         readonly format: "text" | "json";
         readonly state: "captured" | "disabled" | "unavailable";
       };
+      readonly startedAt: number;
+      readonly index: number;
       readonly error: {
         readonly text: string;
         readonly truncated: boolean;
@@ -434,14 +434,9 @@ declare const EvalCell: Schema.Struct<{
       sha256: Schema.filter<typeof Schema.String>;
     }>>>;
     validations: Schema.optional<Schema.refine<readonly {
-      readonly name: string;
-      readonly id: string;
-      readonly kind: "command" | "judge" | "code";
-      readonly status: "passed" | "running" | "failed" | "error" | "queued" | "skipped";
+      readonly kind: "judge" | "command" | "code";
       readonly message: string;
-      readonly startedAt: number | null;
-      readonly index: number;
-      readonly exitCode: number | null;
+      readonly name: string;
       readonly output: {
         readonly text: string;
         readonly truncated: boolean;
@@ -454,6 +449,11 @@ declare const EvalCell: Schema.Struct<{
         readonly format: "text" | "json";
         readonly state: "captured" | "disabled" | "unavailable";
       };
+      readonly id: string;
+      readonly status: "passed" | "running" | "failed" | "error" | "queued" | "skipped";
+      readonly startedAt: number | null;
+      readonly index: number;
+      readonly exitCode: number | null;
       readonly error: {
         readonly text: string;
         readonly truncated: boolean;
@@ -480,8 +480,6 @@ declare const EvalCell: Schema.Struct<{
         readonly state: "captured" | "disabled" | "unavailable";
       } | undefined;
       readonly calls: readonly {
-        readonly startedAt: number;
-        readonly index: number;
         readonly output: {
           readonly text: string;
           readonly truncated: boolean;
@@ -494,6 +492,8 @@ declare const EvalCell: Schema.Struct<{
           readonly format: "text" | "json";
           readonly state: "captured" | "disabled" | "unavailable";
         };
+        readonly startedAt: number;
+        readonly index: number;
         readonly error: {
           readonly text: string;
           readonly truncated: boolean;
@@ -758,14 +758,9 @@ declare const EvalRun: Schema.Struct<{
         sha256: Schema.filter<typeof Schema.String>;
       }>>>;
       validations: Schema.optional<Schema.refine<readonly {
-        readonly name: string;
-        readonly id: string;
-        readonly kind: "command" | "judge" | "code";
-        readonly status: "passed" | "running" | "failed" | "error" | "queued" | "skipped";
+        readonly kind: "judge" | "command" | "code";
         readonly message: string;
-        readonly startedAt: number | null;
-        readonly index: number;
-        readonly exitCode: number | null;
+        readonly name: string;
         readonly output: {
           readonly text: string;
           readonly truncated: boolean;
@@ -778,6 +773,11 @@ declare const EvalRun: Schema.Struct<{
           readonly format: "text" | "json";
           readonly state: "captured" | "disabled" | "unavailable";
         };
+        readonly id: string;
+        readonly status: "passed" | "running" | "failed" | "error" | "queued" | "skipped";
+        readonly startedAt: number | null;
+        readonly index: number;
+        readonly exitCode: number | null;
         readonly error: {
           readonly text: string;
           readonly truncated: boolean;
@@ -804,8 +804,6 @@ declare const EvalRun: Schema.Struct<{
           readonly state: "captured" | "disabled" | "unavailable";
         } | undefined;
         readonly calls: readonly {
-          readonly startedAt: number;
-          readonly index: number;
           readonly output: {
             readonly text: string;
             readonly truncated: boolean;
@@ -818,6 +816,8 @@ declare const EvalRun: Schema.Struct<{
             readonly format: "text" | "json";
             readonly state: "captured" | "disabled" | "unavailable";
           };
+          readonly startedAt: number;
+          readonly index: number;
           readonly error: {
             readonly text: string;
             readonly truncated: boolean;
@@ -1108,14 +1108,9 @@ declare const EvalCellHistoryEntry: Schema.Struct<{
       sha256: Schema.filter<typeof Schema.String>;
     }>>>;
     validations: Schema.optional<Schema.refine<readonly {
-      readonly name: string;
-      readonly id: string;
-      readonly kind: "command" | "judge" | "code";
-      readonly status: "passed" | "running" | "failed" | "error" | "queued" | "skipped";
+      readonly kind: "judge" | "command" | "code";
       readonly message: string;
-      readonly startedAt: number | null;
-      readonly index: number;
-      readonly exitCode: number | null;
+      readonly name: string;
       readonly output: {
         readonly text: string;
         readonly truncated: boolean;
@@ -1128,6 +1123,11 @@ declare const EvalCellHistoryEntry: Schema.Struct<{
         readonly format: "text" | "json";
         readonly state: "captured" | "disabled" | "unavailable";
       };
+      readonly id: string;
+      readonly status: "passed" | "running" | "failed" | "error" | "queued" | "skipped";
+      readonly startedAt: number | null;
+      readonly index: number;
+      readonly exitCode: number | null;
       readonly error: {
         readonly text: string;
         readonly truncated: boolean;
@@ -1154,8 +1154,6 @@ declare const EvalCellHistoryEntry: Schema.Struct<{
         readonly state: "captured" | "disabled" | "unavailable";
       } | undefined;
       readonly calls: readonly {
-        readonly startedAt: number;
-        readonly index: number;
         readonly output: {
           readonly text: string;
           readonly truncated: boolean;
@@ -1168,6 +1166,8 @@ declare const EvalCellHistoryEntry: Schema.Struct<{
           readonly format: "text" | "json";
           readonly state: "captured" | "disabled" | "unavailable";
         };
+        readonly startedAt: number;
+        readonly index: number;
         readonly error: {
           readonly text: string;
           readonly truncated: boolean;

package/dist/{evals-Dw2Ek-mx.d.mts → evals-M2G4zvnS.d.mts}

@@ -91,14 +91,9 @@ declare const EvalTrial: Schema.Struct<{
     sha256: Schema.filter<typeof Schema.String>;
   }>>>;
   validations: Schema.optional<Schema.refine<readonly {
-    readonly name: string;
-    readonly id: string;
-    readonly kind: "command" | "judge" | "code";
-    readonly status: "passed" | "running" | "failed" | "error" | "queued" | "skipped";
+    readonly kind: "judge" | "command" | "code";
     readonly message: string;
-    readonly startedAt: number | null;
-    readonly index: number;
-    readonly exitCode: number | null;
+    readonly name: string;
     readonly output: {
       readonly text: string;
       readonly truncated: boolean;
@@ -111,6 +106,11 @@ declare const EvalTrial: Schema.Struct<{
       readonly format: "text" | "json";
       readonly state: "captured" | "disabled" | "unavailable";
     };
+    readonly id: string;
+    readonly status: "passed" | "running" | "failed" | "error" | "queued" | "skipped";
+    readonly startedAt: number | null;
+    readonly index: number;
+    readonly exitCode: number | null;
     readonly error: {
       readonly text: string;
       readonly truncated: boolean;
@@ -137,8 +137,6 @@ declare const EvalTrial: Schema.Struct<{
       readonly state: "captured" | "disabled" | "unavailable";
     } | undefined;
     readonly calls: readonly {
-      readonly startedAt: number;
-      readonly index: number;
       readonly output: {
         readonly text: string;
         readonly truncated: boolean;
@@ -151,6 +149,8 @@ declare const EvalTrial: Schema.Struct<{
         readonly format: "text" | "json";
         readonly state: "captured" | "disabled" | "unavailable";
       };
+      readonly startedAt: number;
+      readonly index: number;
       readonly error: {
         readonly text: string;
         readonly truncated: boolean;
@@ -434,14 +434,9 @@ declare const EvalCell: Schema.Struct<{
       sha256: Schema.filter<typeof Schema.String>;
     }>>>;
     validations: Schema.optional<Schema.refine<readonly {
-      readonly name: string;
-      readonly id: string;
-      readonly kind: "command" | "judge" | "code";
-      readonly status: "passed" | "running" | "failed" | "error" | "queued" | "skipped";
+      readonly kind: "judge" | "command" | "code";
       readonly message: string;
-      readonly startedAt: number | null;
-      readonly index: number;
-      readonly exitCode: number | null;
+      readonly name: string;
       readonly output: {
         readonly text: string;
         readonly truncated: boolean;
@@ -454,6 +449,11 @@ declare const EvalCell: Schema.Struct<{
         readonly format: "text" | "json";
         readonly state: "captured" | "disabled" | "unavailable";
       };
+      readonly id: string;
+      readonly status: "passed" | "running" | "failed" | "error" | "queued" | "skipped";
+      readonly startedAt: number | null;
+      readonly index: number;
+      readonly exitCode: number | null;
       readonly error: {
         readonly text: string;
         readonly truncated: boolean;
@@ -480,8 +480,6 @@ declare const EvalCell: Schema.Struct<{
         readonly state: "captured" | "disabled" | "unavailable";
       } | undefined;
       readonly calls: readonly {
-        readonly startedAt: number;
-        readonly index: number;
         readonly output: {
           readonly text: string;
           readonly truncated: boolean;
@@ -494,6 +492,8 @@ declare const EvalCell: Schema.Struct<{
           readonly format: "text" | "json";
           readonly state: "captured" | "disabled" | "unavailable";
         };
+        readonly startedAt: number;
+        readonly index: number;
         readonly error: {
           readonly text: string;
           readonly truncated: boolean;
@@ -758,14 +758,9 @@ declare const EvalRun: Schema.Struct<{
         sha256: Schema.filter<typeof Schema.String>;
       }>>>;
       validations: Schema.optional<Schema.refine<readonly {
-        readonly name: string;
-        readonly id: string;
-        readonly kind: "command" | "judge" | "code";
-        readonly status: "passed" | "running" | "failed" | "error" | "queued" | "skipped";
+        readonly kind: "judge" | "command" | "code";
         readonly message: string;
-        readonly startedAt: number | null;
-        readonly index: number;
-        readonly exitCode: number | null;
+        readonly name: string;
         readonly output: {
           readonly text: string;
           readonly truncated: boolean;
@@ -778,6 +773,11 @@ declare const EvalRun: Schema.Struct<{
           readonly format: "text" | "json";
           readonly state: "captured" | "disabled" | "unavailable";
         };
+        readonly id: string;
+        readonly status: "passed" | "running" | "failed" | "error" | "queued" | "skipped";
+        readonly startedAt: number | null;
+        readonly index: number;
+        readonly exitCode: number | null;
         readonly error: {
           readonly text: string;
           readonly truncated: boolean;
@@ -804,8 +804,6 @@ declare const EvalRun: Schema.Struct<{
           readonly state: "captured" | "disabled" | "unavailable";
         } | undefined;
         readonly calls: readonly {
-          readonly startedAt: number;
-          readonly index: number;
           readonly output: {
             readonly text: string;
             readonly truncated: boolean;
@@ -818,6 +816,8 @@ declare const EvalRun: Schema.Struct<{
             readonly format: "text" | "json";
             readonly state: "captured" | "disabled" | "unavailable";
           };
+          readonly startedAt: number;
+          readonly index: number;
           readonly error: {
             readonly text: string;
             readonly truncated: boolean;
@@ -1108,14 +1108,9 @@ declare const EvalCellHistoryEntry: Schema.Struct<{
       sha256: Schema.filter<typeof Schema.String>;
     }>>>;
     validations: Schema.optional<Schema.refine<readonly {
-      readonly name: string;
-      readonly id: string;
-      readonly kind: "command" | "judge" | "code";
-      readonly status: "passed" | "running" | "failed" | "error" | "queued" | "skipped";
+      readonly kind: "judge" | "command" | "code";
       readonly message: string;
-      readonly startedAt: number | null;
-      readonly index: number;
-      readonly exitCode: number | null;
+      readonly name: string;
       readonly output: {
         readonly text: string;
         readonly truncated: boolean;
@@ -1128,6 +1123,11 @@ declare const EvalCellHistoryEntry: Schema.Struct<{
         readonly format: "text" | "json";
         readonly state: "captured" | "disabled" | "unavailable";
       };
+      readonly id: string;
+      readonly status: "passed" | "running" | "failed" | "error" | "queued" | "skipped";
+      readonly startedAt: number | null;
+      readonly index: number;
+      readonly exitCode: number | null;
       readonly error: {
         readonly text: string;
         readonly truncated: boolean;
@@ -1154,8 +1154,6 @@ declare const EvalCellHistoryEntry: Schema.Struct<{
         readonly state: "captured" | "disabled" | "unavailable";
       } | undefined;
       readonly calls: readonly {
-        readonly startedAt: number;
-        readonly index: number;
         readonly output: {
           readonly text: string;
           readonly truncated: boolean;
@@ -1168,6 +1166,8 @@ declare const EvalCellHistoryEntry: Schema.Struct<{
           readonly format: "text" | "json";
           readonly state: "captured" | "disabled" | "unavailable";
         };
+        readonly startedAt: number;
+        readonly index: number;
         readonly error: {
           readonly text: string;
           readonly truncated: boolean;

package/dist/{evals-api-CNoLlXL5.cjs → evals-api-DUiZWsyC.cjs}

@@ -117,6 +117,8 @@ var ApiKeyAuthentication = class extends _effect_platform.HttpApiMiddleware.Tag(
 const PROMPT_LIMIT = 16384;
 const VERIFY_LIMIT = 8192;
 const NAME_LIMIT = 200;
+const TAG_LIMIT = 40;
+const TAGS_PER_CASE = 12;
 const VARIABLE_VALUE_LIMIT = 2048;
 const EvalPrompt = effect.Schema.String.pipe(effect.Schema.maxLength(PROMPT_LIMIT), effect.Schema.annotations({
 	description: "What the agent is asked to do.",
@@ -133,6 +135,11 @@ const EvalCaseName = effect.Schema.String.pipe(effect.Schema.maxLength(NAME_LIMI
 	identifier: "EvalCaseName",
 	message: () => `A case name must be at most ${NAME_LIMIT} characters.`
 }));
+effect.Schema.Array(effect.Schema.String.pipe(effect.Schema.minLength(1), effect.Schema.maxLength(TAG_LIMIT))).pipe(effect.Schema.maxItems(TAGS_PER_CASE), effect.Schema.annotations({
+	description: "What a case is grouped under.",
+	identifier: "EvalCaseTags",
+	message: () => `A case may carry at most ${TAGS_PER_CASE} tags of ${TAG_LIMIT} characters.`
+}));
 const EvalVariableValue = effect.Schema.String.pipe(effect.Schema.maxLength(VARIABLE_VALUE_LIMIT), effect.Schema.annotations({
 	identifier: "EvalVariableValue",
 	message: () => `A variable value must be at most ${VARIABLE_VALUE_LIMIT} characters, because it is substituted into the prompt before it is quoted.`
@@ -536,6 +543,28 @@ const EvalRunPage = effect.Schema.Struct({
 	runs: effect.Schema.Array(EvalRunSummary),
 	total: effect.Schema.Int
 });
+const EvalCaseSummary = effect.Schema.Struct({
+	cellKey: effect.Schema.String,
+	distribution: EvalDistribution,
+	harness: effect.Schema.String,
+	lastRunAtMillis: effect.Schema.Int,
+	lastRunId: effect.Schema.String,
+	model: effect.Schema.String,
+	name: effect.Schema.String,
+	runCount: effect.Schema.Int,
+	suite: effect.Schema.NullOr(effect.Schema.String),
+	tags: effect.Schema.Array(effect.Schema.String)
+}).annotations({
+	description: "A case as the list shows it, with its newest run.",
+	identifier: "EvalCaseSummary"
+});
+const EvalCasePage = effect.Schema.Struct({
+	cases: effect.Schema.Array(EvalCaseSummary),
+	tags: effect.Schema.Array(effect.Schema.String)
+}).annotations({
+	description: "Cases and every tag they carry between them.",
+	identifier: "EvalCasePage"
+});
 const EvalCellHistoryEntry = effect.Schema.Struct({
 	trigger: effect.Schema.optionalWith(effect.Schema.NullOr(EvalTrigger), { default: () => null }),
 	distribution: EvalDistribution,
@@ -727,6 +756,13 @@ const PublicEvalSandbox = effect.Schema.Literal(...HOSTED_SANDBOXES).annotations
 	description: "The hosted sandbox a task runs in.",
 	identifier: "PublicEvalSandbox"
 });
+const ListCasesRequest = effect.Schema.Struct({
+	limit: effect.Schema.optional(effect.Schema.Int),
+	tag: effect.Schema.optional(effect.Schema.NullOr(effect.Schema.String))
+}).annotations({
+	description: "Which cases to read.",
+	identifier: "ListCasesRequest"
+});
 const ListEvalsRequest = effect.Schema.Struct({
 	cursor: effect.Schema.optional(effect.Schema.NullOr(EvalPageCursor)),
 	limit: effect.Schema.optional(effect.Schema.Int)
@@ -768,6 +804,23 @@ const PublicStartEvalRequest = effect.Schema.Struct({
 	description: `Start a grid with at most 100 total case, task, and trial combinations.`,
 	identifier: "StartEvalRequest"
 });
+const CredentialLeaseRequest = effect.Schema.Struct({
+	harness: EvalHarness,
+	id: effect.Schema.String
+}).annotations({
+	description: "Ask for the credentials a run the caller is executing needs, for the harness it names.",
+	identifier: "CredentialLeaseRequest"
+});
+const CredentialLease = effect.Schema.Struct({
+	expiresAt: effect.Schema.DateTimeUtc,
+	values: effect.Schema.Record({
+		key: effect.Schema.String,
+		value: effect.Schema.String
+	})
+}).annotations({
+	description: "Credentials for one run, held in memory and never written down. Short-lived: start another run rather than keeping these.",
+	identifier: "CredentialLease"
+});
 const ReportedTrial = effect.Schema.Struct({
 	caseName: EvalCaseName,
 	events: effect.Schema.Array(HarnessEvent),
@@ -787,7 +840,7 @@ const ReportTrialRequest = effect.Schema.Struct({
 	description: "A trial the caller ran, named by the run it belongs to.",
 	identifier: "ReportTrialRequest"
 });
-var PublicEvalsGroup = class extends _effect_platform.HttpApiGroup.make("evals").add(_effect_platform.HttpApiEndpoint.post("list", "/evals.list").setPayload(ListEvalsRequest).addSuccess(EvalRunPage).annotate(_effect_platform.OpenApi.Summary, "List eval runs").annotate(_effect_platform.OpenApi.Description, "Newest first. Pass the `next` cursor from a response to read the page after it; a null `next` means there are no more.")).add(_effect_platform.HttpApiEndpoint.post("start", "/evals.start").setPayload(PublicStartEvalRequest).addSuccess(StartedEval).annotate(_effect_platform.OpenApi.Summary, "Start an eval run").annotate(_effect_platform.OpenApi.Description, `Starts the grid and returns its id while trials continue in the background. ${PROFILE_HARNESS_RULE}`)).add(_effect_platform.HttpApiEndpoint.post("reportTrial", "/evals.reportTrial").setPayload(ReportTrialRequest).addSuccess(effect.Schema.Void).annotate(_effect_platform.OpenApi.Summary, "Report a trial run outside the platform").annotate(_effect_platform.OpenApi.Description, "For a run started with executeLocally. The result is recorded as reported: it was produced somewhere the platform cannot inspect, so it is marked and kept out of baselines.")).add(_effect_platform.HttpApiEndpoint.post("finishRun", "/evals.finishRun").setPayload(EvalRunRequest).addSuccess(EvalRun).annotate(_effect_platform.OpenApi.Summary, "Settle a run whose trials the caller ran").annotate(_effect_platform.OpenApi.Description, "Settles the run from the trials reported so far. A run left unsettled is swept like any other abandoned work.")).add(_effect_platform.HttpApiEndpoint.post("artifact", "/evals.artifact").setPayload(EvalArtifactRequest).addSuccess(EvalArtifact)).add(_effect_platform.HttpApiEndpoint.post("get", "/evals.get").setPayload(EvalRunRequest).addSuccess(EvalRun).annotate(_effect_platform.OpenApi.Summary, "Get an eval run")).add(_effect_platform.HttpApiEndpoint.post("subscription", "/evals.subscription").setPayload(EvalRunRequest).addSuccess(RunSubscription).annotate(_effect_platform.OpenApi.Summary, "Watch an eval run as it moves").annotate(_effect_platform.OpenApi.Description, "Returns a read-only token scoped to this run, for following it in real time. Short-lived: request another when it expires.")).add(_effect_platform.HttpApiEndpoint.post("cellHistory", "/evals.cellHistory").setPayload(EvalCellRequest).addSuccess(effect.Schema.Array(EvalCellHistoryEntry)).annotate(_effect_platform.OpenApi.Summary, "List a cell's history").annotate(_effect_platform.OpenApi.Description, "Returns the 20 most recent results.")).add(_effect_platform.HttpApiEndpoint.post("rerunCell", "/evals.rerunCell").setPayload(effect.Schema.extend(EvalRunRequest, effect.Schema.extend(EvalCellRequest, RerunCellRequest))).addSuccess(StartedEval).annotate(_effect_platform.OpenApi.Summary, "Rerun one cell")).add(_effect_platform.HttpApiEndpoint.post("models", "/evals.models").setPayload(EvalModelsRequest).addSuccess(ModelCatalogue).annotate(_effect_platform.OpenApi.Summary, "List models available to the harness").annotate(_effect_platform.OpenApi.Description, "The command harness has no catalogue of its own, so its list is empty: the model is whatever the profile's run command reads from ANPORD_MODEL.")).addError(BadRequest).addError(Conflict).addError(Forbidden).addError(NotFound).middleware(ApiKeyAuthentication).annotate(_effect_platform.OpenApi.Title, "Evals").annotate(_effect_platform.OpenApi.Description, "Run cases across harness, model, and sandbox combinations and compare the results with their baselines.") {};
+var PublicEvalsGroup = class extends _effect_platform.HttpApiGroup.make("evals").add(_effect_platform.HttpApiEndpoint.post("list", "/evals.list").setPayload(ListEvalsRequest).addSuccess(EvalRunPage).annotate(_effect_platform.OpenApi.Summary, "List eval runs").annotate(_effect_platform.OpenApi.Description, "Newest first. Pass the `next` cursor from a response to read the page after it; a null `next` means there are no more.")).add(_effect_platform.HttpApiEndpoint.post("cases", "/evals.cases").setPayload(ListCasesRequest).addSuccess(EvalCasePage).annotate(_effect_platform.OpenApi.Summary, "List cases").annotate(_effect_platform.OpenApi.Description, "One row per case, carrying its newest run. Pass `tag` to read only the cases carrying it.")).add(_effect_platform.HttpApiEndpoint.post("start", "/evals.start").setPayload(PublicStartEvalRequest).addSuccess(StartedEval).annotate(_effect_platform.OpenApi.Summary, "Start an eval run").annotate(_effect_platform.OpenApi.Description, `Starts the grid and returns its id while trials continue in the background. ${PROFILE_HARNESS_RULE}`)).add(_effect_platform.HttpApiEndpoint.post("credentials", "/evals.credentials").setPayload(CredentialLeaseRequest).addSuccess(CredentialLease).annotate(_effect_platform.OpenApi.Summary, "Lease the credentials a local run needs").annotate(_effect_platform.OpenApi.Description, "For a run started with executeLocally, so the machine running it holds no credentials of its own. Returns the organization credential for one harness, expiring in minutes. Sandbox credentials are never leased: a local run opens no cloud sandbox.")).add(_effect_platform.HttpApiEndpoint.post("reportTrial", "/evals.reportTrial").setPayload(ReportTrialRequest).addSuccess(effect.Schema.Void).annotate(_effect_platform.OpenApi.Summary, "Report a trial run outside the platform").annotate(_effect_platform.OpenApi.Description, "For a run started with executeLocally. The result is recorded as reported: it was produced somewhere the platform cannot inspect, so it is marked and kept out of baselines.")).add(_effect_platform.HttpApiEndpoint.post("finishRun", "/evals.finishRun").setPayload(EvalRunRequest).addSuccess(EvalRun).annotate(_effect_platform.OpenApi.Summary, "Settle a run whose trials the caller ran").annotate(_effect_platform.OpenApi.Description, "Settles the run from the trials reported so far. A run left unsettled is swept like any other abandoned work.")).add(_effect_platform.HttpApiEndpoint.post("artifact", "/evals.artifact").setPayload(EvalArtifactRequest).addSuccess(EvalArtifact)).add(_effect_platform.HttpApiEndpoint.post("get", "/evals.get").setPayload(EvalRunRequest).addSuccess(EvalRun).annotate(_effect_platform.OpenApi.Summary, "Get an eval run")).add(_effect_platform.HttpApiEndpoint.post("subscription", "/evals.subscription").setPayload(EvalRunRequest).addSuccess(RunSubscription).annotate(_effect_platform.OpenApi.Summary, "Watch an eval run as it moves").annotate(_effect_platform.OpenApi.Description, "Returns a read-only token scoped to this run, for following it in real time. Short-lived: request another when it expires.")).add(_effect_platform.HttpApiEndpoint.post("cellHistory", "/evals.cellHistory").setPayload(EvalCellRequest).addSuccess(effect.Schema.Array(EvalCellHistoryEntry)).annotate(_effect_platform.OpenApi.Summary, "List a cell's history").annotate(_effect_platform.OpenApi.Description, "Returns the 20 most recent results.")).add(_effect_platform.HttpApiEndpoint.post("rerunCell", "/evals.rerunCell").setPayload(effect.Schema.extend(EvalRunRequest, effect.Schema.extend(EvalCellRequest, RerunCellRequest))).addSuccess(StartedEval).annotate(_effect_platform.OpenApi.Summary, "Rerun one cell")).add(_effect_platform.HttpApiEndpoint.post("models", "/evals.models").setPayload(EvalModelsRequest).addSuccess(ModelCatalogue).annotate(_effect_platform.OpenApi.Summary, "List models available to the harness").annotate(_effect_platform.OpenApi.Description, "The command harness has no catalogue of its own, so its list is empty: the model is whatever the profile's run command reads from ANPORD_MODEL.")).addError(BadRequest).addError(Conflict).addError(Forbidden).addError(NotFound).middleware(ApiKeyAuthentication).annotate(_effect_platform.OpenApi.Title, "Evals").annotate(_effect_platform.OpenApi.Description, "Run cases across harness, model, and sandbox combinations and compare the results with their baselines.") {};
 //#endregion
 Object.defineProperty(exports, "ApiKeyAuthentication", {
 	enumerable: true,