Use the configured Runner default in normalization tests

The upstream Codex default update left two server test expectations on
the previous model. Compare normalization with the shared default so a
merge with current master checks the active policy. Explicit model
preservation still has its separate assertion.

All six cases pass with both the branch default and the current upstream
default. The prior CI failure is recorded in the PR.

Co-Authored-By: Paperclip <noreply@paperclip.ing>

server/src/__tests__/agents-paperclip-runner-skills.test.ts

@@ -1,4 +1,5 @@
 import { describe, expect, it } from "vitest";
+import { PAPERCLIP_RUNNER_DEFAULT_MODELS } from "@paperclipai/adapter-utils/paperclip-runner-permissions";
 import {
   normalizePaperclipOperationalSkillPreference,
   normalizePaperclipRunnerAdapterConfig,
@@ -16,7 +17,7 @@ describe("paperclip_runner operational skill normalization", () => {
   it("applies full-auto native runner defaults at persistence boundaries", () => {
     expect(normalizePaperclipRunnerAdapterConfig("paperclip_runner", {})).toEqual({
       provider: "codex",
-      model: "gpt-5.6-sol",
+      model: PAPERCLIP_RUNNER_DEFAULT_MODELS.codex,
       codexPermissionMode: "never",
       lifecycleMode: "per_turn",
     });
@@ -24,7 +25,7 @@ describe("paperclip_runner operational skill normalization", () => {
 
   it("repairs an existing blank model without replacing an explicit model", () => {
     expect(normalizePaperclipRunnerAdapterConfig("paperclip_runner", { model: "" }))
-      .toMatchObject({ model: "gpt-5.6-sol" });
+      .toMatchObject({ model: PAPERCLIP_RUNNER_DEFAULT_MODELS.codex });
     expect(normalizePaperclipRunnerAdapterConfig("paperclip_runner", { model: "gpt-5.5" }))
       .toMatchObject({ model: "gpt-5.5" });
   });