@tangle-network/agent-eval 0.120.4 → 0.122.0

package/dist/openapi.json

@@ -2,7 +2,7 @@
   "openapi": "3.1.0",
   "info": {
     "title": "@tangle-network/agent-eval — wire protocol",
-    "version": "0.120.4",
+    "version": "0.122.0",
     "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/dist/rl.d.ts

@@ -3700,11 +3700,15 @@ interface GenerationCandidate {
      *  reflective proposer grounds its next edit on. Keep `notes` GENERALIZABLE
      *  (which checks/lines/dimensions failed and how), NOT case-specific ground
      *  truth: leaking expected answers into the prompt is memorization, and the
-     *  held-out gate would reject it anyway. */
+     *  held-out gate would reject it anyway. `emitted` is a bounded excerpt of
+     *  the candidate's raw output for the scenario (worst rep when reps > 1) —
+     *  the "what it actually did" evidence; optional so trajectory capture is
+     *  never required of a dispatch. */
     scenarios: Array<{
         scenarioId: string;
         composite: number;
         notes?: string;
+        emitted?: string;
     }>;
     /** Proposer-supplied short label for the change. Present when the proposer
      *  returned a `ProposedCandidate`; absent for bare-surface mutators. */

package/dist/{run-campaign-75RTPVV5.js → run-campaign-HNFPJET4.js}

@@ -1,9 +1,9 @@
 import {
   planCampaignRun,
   runCampaign
-} from "./chunk-SYV364BL.js";
+} from "./chunk-XDIRG3TO.js";
 import "./chunk-PJQFMIOX.js";
-import "./chunk-ZMXDQ4K7.js";
+import "./chunk-JHOJHHU7.js";
 import "./chunk-VI2UW6B6.js";
 import "./chunk-ONWEPEDO.js";
 import "./chunk-PZ5AY32C.js";
@@ -11,4 +11,4 @@ export {
   planCampaignRun,
   runCampaign
 };
-//# sourceMappingURL=run-campaign-75RTPVV5.js.map
+//# sourceMappingURL=run-campaign-HNFPJET4.js.map

package/dist/wire/index.js

@@ -34,9 +34,9 @@ import {
   runRpcOnce,
   startServer,
   startServerAsync
-} from "../chunk-NMN4WGSJ.js";
+} from "../chunk-4I2E3LLO.js";
 import "../chunk-NJC7U437.js";
-import "../chunk-ZMXDQ4K7.js";
+import "../chunk-JHOJHHU7.js";
 import "../chunk-VI2UW6B6.js";
 import "../chunk-PC4UYEBM.js";
 import "../chunk-ONWEPEDO.js";

package/docs/eval-surface-map.md

@@ -12,7 +12,7 @@ composition point already exists (see Produced-state grading below).**
 | `runCampaign` | The measurement primitive. Run a dispatch over scenarios × seeds × reps, score each with judges, aggregate. Caller owns the dispatch. | `CampaignResult` |
 | `runEval` | The simplest preset over `runCampaign` — just score, no loop, no gate. The 80% "I want a scorecard" case. | `CampaignResult` |
 | `runProfileMatrix` | Factor the SAME scenarios across N agent **profiles** (models / prompt variants), with RunRecord stamping + a real-backend integrity guard. | `RunRecord[]` |
-| `runOptimization` | GENERATE: measured or validated premeasured baseline → N generations of propose → measure → rank → promote. No release gate. | generations + winner |
+| `runOptimization` | GENERATE: baseline → N generations of propose → measure → rank → promote. No release gate. | generations + winner |
 | `runImprovementLoop` | The release-gate shell around `runOptimization`: adds a held-out re-score + a promotion gate (+ optional auto-PR). | gate decision + winner |
 | `runEvalCampaign` | Inversion-of-control variant of `runCampaign` — the runner is handed a pre-wired trace/sink/emitter and integrity gating as a precondition. Use when you need full capture by construction. | `CampaignResult` + records |
 | `runAgentMatrix` | The bare N-axis cartesian scheduler with concurrency control. The layer beneath the eval surface — reach for it only when you need raw scheduling, not eval semantics. | cell results |

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@tangle-network/agent-eval",
-  "version": "0.120.4",
+  "version": "0.122.0",
   "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": {
@@ -152,7 +152,7 @@
     "@ax-llm/ax": "^23.0.0",
     "@hono/node-server": "^2.0.0",
     "@tangle-network/agent-core": "^0.4.11",
-    "@tangle-network/agent-interface": "^0.26.0",
+    "@tangle-network/agent-interface": "^0.29.0",
     "@tangle-network/tcloud": "^0.4.14",
     "hono": "^4.12.30",
     "zod": "^4.3.6"