openpond-sdk 0.5.13 → 0.5.15

package/TASKSETS.md

@@ -245,3 +245,16 @@ Model revisions, publish/select the package, finalize the draft and retain the
 retry receipt. The client verifies response scope, revisions and file hashes;
 it never retries writes automatically. Hosts own authorization, durable storage,
 pagination, transaction isolation and idempotent recovery.
+
+### Publish a Model with retained evaluation
+
+When synchronizing a local Model, publish its complete evaluation Taskset alongside
+its training package using `evaluationPackage` and `modelConfiguration`. Set
+`modelConfiguration.trainingSetup.evaluationTasksetRef` to the evaluation package's
+portable Taskset release identity, not the local Taskset identity. The publication
+receipt includes `evaluation.taskset`, `evaluation.packageHash` and
+`evaluation.hostedTasksetId`; the SDK rejects a missing or mismatched receipt.
+
+The same configuration carries the saved recipe and limits. Publication does not
+run training or prove evaluation isolation, grader quality or training readiness.
+An attachment-only publication cannot change Model configuration.

package/TRAINING_PROTOCOL.md

@@ -204,3 +204,9 @@ An explicitly selected Harness continues to use its verified source package.
 The compiler preserves existing release hashes and rejects dependencies whose
 visibility is not `policy`; environment, task tools and graders remain bound by
 the Taskset and training bundle.
+
+## Published Harness source
+
+`createTrainingClient().publishHarnessSource(sourcePackage)` publishes a verified, portable, policy-visible `@openpond/harness` source package to the authenticated workspace without starting a Job. `getHarnessSource({ id, contentHash })` retrieves and verifies that exact release. The endpoints are `PUT` and `GET /v1/training/harness-sources/:releaseId/:contentHash`; they use the existing Training media type and read/write scopes.
+
+Hosted learning resolves this immutable source with Desktop closed. Publish before selecting the release on a hosted Model. Preparation still checks the Taskset adapter, required tools, private grading boundary and effective context; publication is not training or quality qualification. Existing prepared Jobs retain their captured source when a newer release is published.