@tangle-network/agent-eval 0.122.9 → 0.123.0

package/dist/index.d.ts

@@ -14252,6 +14252,10 @@ interface RunImprovementLoopResult<TArtifact, TScenario extends Scenario> extend
     neutralizedOnHoldout?: CampaignResult<TArtifact, TScenario>;
     neutralizedSurface?: MutableSurface;
     gateResult: Awaited<ReturnType<Gate<TArtifact, TScenario>['decide']>>;
+    /** Present iff the loop ran with `holdout: 'deferred'`. When set,
+     *  `baselineOnHoldout`/`winnerOnHoldout` are the shared EMPTY campaign (zero
+     *  cells dispatched) and the gate verdict is the forced `'hold'`. */
+    holdout?: 'deferred';
     /** Unified baseline→winner surface diff. Computed UNCONDITIONALLY (not only
      *  when `autoOnPromote === 'pr'`) so the diff that the gate decided on is
      *  always present on the result + in the emitted provenance record. Empty

package/dist/index.js

@@ -53,7 +53,7 @@ import {
   pairArms,
   parseCorrectnessResponse,
   verifyCompletion
-} from "./chunk-CWF6SCBK.js";
+} from "./chunk-GS3FJGUF.js";
 import {
   DEFAULT_MUTATION_PRIMITIVES,
   DEFAULT_RED_TEAM_CORPUS,
@@ -91,7 +91,7 @@ import {
   scoreRedTeamOutput,
   surfaceContentHash,
   toolNamesForRun
-} from "./chunk-4EDDIK77.js";
+} from "./chunk-HZJF4IUO.js";
 import {
   BackendIntegrityError,
   assertRealAgentReceipts,

package/dist/openapi.json

@@ -2,7 +2,7 @@
   "openapi": "3.1.0",
   "info": {
     "title": "@tangle-network/agent-eval — wire protocol",
-    "version": "0.122.9",
+    "version": "0.123.0",
     "description": "HTTP and stdio RPC interface to agent-eval. The TypeScript runtime is the source of truth; this spec is the contract that cross-language clients (Python, Rust, Go) generate from.\n\nWire-protocol version: 1.0.0. Bumps on breaking changes to request/response schemas.",
     "contact": {
       "name": "Tangle Network",

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@tangle-network/agent-eval",
-  "version": "0.122.9",
+  "version": "0.123.0",
   "description": "Evaluate and improve AI agents from runs, traces, judges, and feedback. Compare candidates, cluster failures, measure lift, and gate releases.",
   "homepage": "https://github.com/tangle-network/agent-eval#readme",
   "repository": {