diff --git a/.github/workflows/e2e-script.yaml b/.github/workflows/e2e-script.yaml
index 8fdef8f6ff..0b729fead7 100644
--- a/.github/workflows/e2e-script.yaml
+++ b/.github/workflows/e2e-script.yaml
@@ -231,6 +231,7 @@ jobs:
             printf 'NEMOCLAW_ENDPOINT_URL=https://inference-api.nvidia.com/v1\n'
             printf 'NEMOCLAW_MODEL=nvidia/nvidia/nemotron-3-super-v3\n'
             printf 'NEMOCLAW_COMPAT_MODEL=nvidia/nvidia/nemotron-3-super-v3\n'
+            printf 'NEMOCLAW_PREFERRED_API=openai-completions\n'
             printf 'COMPATIBLE_API_KEY=%s\n' "${NVIDIA_INFERENCE_API_KEY}"
           } >> "$GITHUB_ENV"
 
diff --git a/.github/workflows/e2e-vitest-scenarios.yaml b/.github/workflows/e2e-vitest-scenarios.yaml
index b8a7c6816a..a77db9295d 100644
--- a/.github/workflows/e2e-vitest-scenarios.yaml
+++ b/.github/workflows/e2e-vitest-scenarios.yaml
@@ -992,6 +992,7 @@ jobs:
           NEMOCLAW_ENDPOINT_URL: https://inference-api.nvidia.com/v1
           NEMOCLAW_MODEL: nvidia/nvidia/nemotron-3-super-v3
           NEMOCLAW_COMPAT_MODEL: nvidia/nvidia/nemotron-3-super-v3
+          NEMOCLAW_PREFERRED_API: openai-completions
         run: |
           set -euo pipefail
           npx vitest run --project e2e-scenarios-live \
diff --git a/.github/workflows/nightly-e2e.yaml b/.github/workflows/nightly-e2e.yaml
index ce97dce38b..98619d67d4 100644
--- a/.github/workflows/nightly-e2e.yaml
+++ b/.github/workflows/nightly-e2e.yaml
@@ -470,6 +470,7 @@ jobs:
           NEMOCLAW_ENDPOINT_URL: https://inference-api.nvidia.com/v1
           NEMOCLAW_MODEL: nvidia/nvidia/nemotron-3-super-v3
           NEMOCLAW_COMPAT_MODEL: nvidia/nvidia/nemotron-3-super-v3
+          NEMOCLAW_PREFERRED_API: openai-completions
           COMPATIBLE_API_KEY: ${{ (github.event_name != 'workflow_dispatch' || inputs.target_ref == '') && secrets.NVIDIA_INFERENCE_API_KEY || '' }}
           NEMOCLAW_NON_INTERACTIVE: "1"
           NEMOCLAW_ACCEPT_THIRD_PARTY_SOFTWARE: "1"
@@ -545,6 +546,7 @@ jobs:
           NEMOCLAW_ENDPOINT_URL: https://inference-api.nvidia.com/v1
           NEMOCLAW_MODEL: nvidia/nvidia/nemotron-3-super-v3
           NEMOCLAW_COMPAT_MODEL: nvidia/nvidia/nemotron-3-super-v3
+          NEMOCLAW_PREFERRED_API: openai-completions
           COMPATIBLE_API_KEY: ${{ (github.event_name != 'workflow_dispatch' || inputs.target_ref == '') && secrets.NVIDIA_INFERENCE_API_KEY || '' }}
           NEMOCLAW_ISSUE_4434_LIVE: "1"
           NEMOCLAW_NON_INTERACTIVE: "1"
@@ -973,6 +975,7 @@ jobs:
           NEMOCLAW_ENDPOINT_URL: https://inference-api.nvidia.com/v1
           NEMOCLAW_MODEL: nvidia/nvidia/nemotron-3-super-v3
           NEMOCLAW_COMPAT_MODEL: nvidia/nvidia/nemotron-3-super-v3
+          NEMOCLAW_PREFERRED_API: openai-completions
           COMPATIBLE_API_KEY: ${{ (github.event_name != 'workflow_dispatch' || inputs.target_ref == '') && secrets.NVIDIA_INFERENCE_API_KEY || '' }}
           NEMOCLAW_NON_INTERACTIVE: "1"
           NEMOCLAW_ACCEPT_THIRD_PARTY_SOFTWARE: "1"
@@ -1270,6 +1273,7 @@ jobs:
           NEMOCLAW_ENDPOINT_URL: https://inference-api.nvidia.com/v1
           NEMOCLAW_MODEL: nvidia/nvidia/nemotron-3-super-v3
           NEMOCLAW_COMPAT_MODEL: nvidia/nvidia/nemotron-3-super-v3
+          NEMOCLAW_PREFERRED_API: openai-completions
           COMPATIBLE_API_KEY: ${{ (github.event_name != 'workflow_dispatch' || inputs.target_ref == '') && secrets.NVIDIA_INFERENCE_API_KEY || '' }}
           NEMOCLAW_NON_INTERACTIVE: "1"
           NEMOCLAW_ACCEPT_THIRD_PARTY_SOFTWARE: "1"
@@ -1576,6 +1580,7 @@ jobs:
           NEMOCLAW_ENDPOINT_URL: https://inference-api.nvidia.com/v1
           NEMOCLAW_MODEL: nvidia/nvidia/nemotron-3-super-v3
           NEMOCLAW_COMPAT_MODEL: nvidia/nvidia/nemotron-3-super-v3
+          NEMOCLAW_PREFERRED_API: openai-completions
           COMPATIBLE_API_KEY: ${{ (github.event_name != 'workflow_dispatch' || inputs.target_ref == '') && secrets.NVIDIA_INFERENCE_API_KEY || '' }}
           E2E_ARTIFACT_DIR: ${{ github.workspace }}/e2e-artifacts/vitest/credential-migration
           NEMOCLAW_RUN_E2E_SCENARIOS: "1"
@@ -1801,6 +1806,7 @@ jobs:
           NEMOCLAW_ENDPOINT_URL: https://inference-api.nvidia.com/v1
           NEMOCLAW_MODEL: nvidia/nvidia/nemotron-3-super-v3
           NEMOCLAW_COMPAT_MODEL: nvidia/nvidia/nemotron-3-super-v3
+          NEMOCLAW_PREFERRED_API: openai-completions
           COMPATIBLE_API_KEY: ${{ (github.event_name != 'workflow_dispatch' || inputs.target_ref == '') && secrets.NVIDIA_INFERENCE_API_KEY || '' }}
           NEMOCLAW_NON_INTERACTIVE: "1"
           NEMOCLAW_ACCEPT_THIRD_PARTY_SOFTWARE: "1"
@@ -1813,6 +1819,7 @@ jobs:
           NEMOCLAW_ENDPOINT_URL: https://inference-api.nvidia.com/v1
           NEMOCLAW_MODEL: nvidia/nvidia/nemotron-3-super-v3
           NEMOCLAW_COMPAT_MODEL: nvidia/nvidia/nemotron-3-super-v3
+          NEMOCLAW_PREFERRED_API: openai-completions
           COMPATIBLE_API_KEY: ${{ (github.event_name != 'workflow_dispatch' || inputs.target_ref == '') && secrets.NVIDIA_INFERENCE_API_KEY || '' }}
           NEMOCLAW_NON_INTERACTIVE: "1"
           NEMOCLAW_ACCEPT_THIRD_PARTY_SOFTWARE: "1"
@@ -1851,6 +1858,7 @@ jobs:
           NEMOCLAW_ENDPOINT_URL: https://inference-api.nvidia.com/v1
           NEMOCLAW_MODEL: nvidia/nvidia/nemotron-3-super-v3
           NEMOCLAW_COMPAT_MODEL: nvidia/nvidia/nemotron-3-super-v3
+          NEMOCLAW_PREFERRED_API: openai-completions
           COMPATIBLE_API_KEY: ${{ (github.event_name != 'workflow_dispatch' || inputs.target_ref == '') && secrets.NVIDIA_INFERENCE_API_KEY || '' }}
           NEMOCLAW_NON_INTERACTIVE: "1"
           NEMOCLAW_ACCEPT_THIRD_PARTY_SOFTWARE: "1"
@@ -1863,6 +1871,7 @@ jobs:
           NEMOCLAW_ENDPOINT_URL: https://inference-api.nvidia.com/v1
           NEMOCLAW_MODEL: nvidia/nvidia/nemotron-3-super-v3
           NEMOCLAW_COMPAT_MODEL: nvidia/nvidia/nemotron-3-super-v3
+          NEMOCLAW_PREFERRED_API: openai-completions
           COMPATIBLE_API_KEY: ${{ (github.event_name != 'workflow_dispatch' || inputs.target_ref == '') && secrets.NVIDIA_INFERENCE_API_KEY || '' }}
           NEMOCLAW_NON_INTERACTIVE: "1"
           NEMOCLAW_ACCEPT_THIRD_PARTY_SOFTWARE: "1"
@@ -1901,6 +1910,7 @@ jobs:
           NEMOCLAW_ENDPOINT_URL: https://inference-api.nvidia.com/v1
           NEMOCLAW_MODEL: nvidia/nvidia/nemotron-3-super-v3
           NEMOCLAW_COMPAT_MODEL: nvidia/nvidia/nemotron-3-super-v3
+          NEMOCLAW_PREFERRED_API: openai-completions
           COMPATIBLE_API_KEY: ${{ (github.event_name != 'workflow_dispatch' || inputs.target_ref == '') && secrets.NVIDIA_INFERENCE_API_KEY || '' }}
           NEMOCLAW_NON_INTERACTIVE: "1"
           NEMOCLAW_ACCEPT_THIRD_PARTY_SOFTWARE: "1"
@@ -1913,6 +1923,7 @@ jobs:
           NEMOCLAW_ENDPOINT_URL: https://inference-api.nvidia.com/v1
           NEMOCLAW_MODEL: nvidia/nvidia/nemotron-3-super-v3
           NEMOCLAW_COMPAT_MODEL: nvidia/nvidia/nemotron-3-super-v3
+          NEMOCLAW_PREFERRED_API: openai-completions
           COMPATIBLE_API_KEY: ${{ (github.event_name != 'workflow_dispatch' || inputs.target_ref == '') && secrets.NVIDIA_INFERENCE_API_KEY || '' }}
           NEMOCLAW_NON_INTERACTIVE: "1"
           NEMOCLAW_ACCEPT_THIRD_PARTY_SOFTWARE: "1"
@@ -1952,6 +1963,7 @@ jobs:
           NEMOCLAW_ENDPOINT_URL: https://inference-api.nvidia.com/v1
           NEMOCLAW_MODEL: nvidia/nvidia/nemotron-3-super-v3
           NEMOCLAW_COMPAT_MODEL: nvidia/nvidia/nemotron-3-super-v3
+          NEMOCLAW_PREFERRED_API: openai-completions
           COMPATIBLE_API_KEY: ${{ (github.event_name != 'workflow_dispatch' || inputs.target_ref == '') && secrets.NVIDIA_INFERENCE_API_KEY || '' }}
           NEMOCLAW_NON_INTERACTIVE: "1"
           NEMOCLAW_ACCEPT_THIRD_PARTY_SOFTWARE: "1"
@@ -1964,6 +1976,7 @@ jobs:
           NEMOCLAW_ENDPOINT_URL: https://inference-api.nvidia.com/v1
           NEMOCLAW_MODEL: nvidia/nvidia/nemotron-3-super-v3
           NEMOCLAW_COMPAT_MODEL: nvidia/nvidia/nemotron-3-super-v3
+          NEMOCLAW_PREFERRED_API: openai-completions
           COMPATIBLE_API_KEY: ${{ (github.event_name != 'workflow_dispatch' || inputs.target_ref == '') && secrets.NVIDIA_INFERENCE_API_KEY || '' }}
           NEMOCLAW_NON_INTERACTIVE: "1"
           NEMOCLAW_ACCEPT_THIRD_PARTY_SOFTWARE: "1"
@@ -2003,6 +2016,7 @@ jobs:
           NEMOCLAW_ENDPOINT_URL: https://inference-api.nvidia.com/v1
           NEMOCLAW_MODEL: nvidia/nvidia/nemotron-3-super-v3
           NEMOCLAW_COMPAT_MODEL: nvidia/nvidia/nemotron-3-super-v3
+          NEMOCLAW_PREFERRED_API: openai-completions
           COMPATIBLE_API_KEY: ${{ (github.event_name != 'workflow_dispatch' || inputs.target_ref == '') && secrets.NVIDIA_INFERENCE_API_KEY || '' }}
           NEMOCLAW_NON_INTERACTIVE: "1"
           NEMOCLAW_ACCEPT_THIRD_PARTY_SOFTWARE: "1"
@@ -2016,6 +2030,7 @@ jobs:
           NEMOCLAW_ENDPOINT_URL: https://inference-api.nvidia.com/v1
           NEMOCLAW_MODEL: nvidia/nvidia/nemotron-3-super-v3
           NEMOCLAW_COMPAT_MODEL: nvidia/nvidia/nemotron-3-super-v3
+          NEMOCLAW_PREFERRED_API: openai-completions
           COMPATIBLE_API_KEY: ${{ (github.event_name != 'workflow_dispatch' || inputs.target_ref == '') && secrets.NVIDIA_INFERENCE_API_KEY || '' }}
           NEMOCLAW_NON_INTERACTIVE: "1"
           NEMOCLAW_ACCEPT_THIRD_PARTY_SOFTWARE: "1"
@@ -2057,6 +2072,7 @@ jobs:
           NEMOCLAW_ENDPOINT_URL: https://inference-api.nvidia.com/v1
           NEMOCLAW_MODEL: nvidia/nvidia/nemotron-3-super-v3
           NEMOCLAW_COMPAT_MODEL: nvidia/nvidia/nemotron-3-super-v3
+          NEMOCLAW_PREFERRED_API: openai-completions
           COMPATIBLE_API_KEY: ${{ (github.event_name != 'workflow_dispatch' || inputs.target_ref == '') && secrets.NVIDIA_INFERENCE_API_KEY || '' }}
           NEMOCLAW_NON_INTERACTIVE: "1"
           NEMOCLAW_ACCEPT_THIRD_PARTY_SOFTWARE: "1"
@@ -2070,6 +2086,7 @@ jobs:
           NEMOCLAW_ENDPOINT_URL: https://inference-api.nvidia.com/v1
           NEMOCLAW_MODEL: nvidia/nvidia/nemotron-3-super-v3
           NEMOCLAW_COMPAT_MODEL: nvidia/nvidia/nemotron-3-super-v3
+          NEMOCLAW_PREFERRED_API: openai-completions
           COMPATIBLE_API_KEY: ${{ (github.event_name != 'workflow_dispatch' || inputs.target_ref == '') && secrets.NVIDIA_INFERENCE_API_KEY || '' }}
           NEMOCLAW_NON_INTERACTIVE: "1"
           NEMOCLAW_ACCEPT_THIRD_PARTY_SOFTWARE: "1"
@@ -2149,6 +2166,7 @@ jobs:
           NEMOCLAW_ENDPOINT_URL: https://inference-api.nvidia.com/v1
           NEMOCLAW_MODEL: nvidia/nvidia/nemotron-3-super-v3
           NEMOCLAW_COMPAT_MODEL: nvidia/nvidia/nemotron-3-super-v3
+          NEMOCLAW_PREFERRED_API: openai-completions
           COMPATIBLE_API_KEY: ${{ (github.event_name != 'workflow_dispatch' || inputs.target_ref == '') && secrets.NVIDIA_INFERENCE_API_KEY || '' }}
           NEMOCLAW_NON_INTERACTIVE: "1"
           NEMOCLAW_ACCEPT_THIRD_PARTY_SOFTWARE: "1"
diff --git a/src/lib/onboard/providers.test.ts b/src/lib/onboard/providers.test.ts
index 56090df090..c7411825f8 100644
--- a/src/lib/onboard/providers.test.ts
+++ b/src/lib/onboard/providers.test.ts
@@ -59,6 +59,7 @@ function withProviderEnv(next: Record<string, string | undefined>, testBody: ()
     "NEMOCLAW_ENDPOINT_URL",
     "NEMOCLAW_MODEL",
     "NEMOCLAW_COMPAT_MODEL",
+    "NEMOCLAW_PREFERRED_API",
     "NEMOCLAW_CLOUD_EXPERIMENTAL_MODEL",
     "NEMOCLAW_E2E_USE_HOSTED_INFERENCE",
     "COMPATIBLE_API_KEY",
@@ -303,11 +304,26 @@ describe("onboard provider helpers", () => {
         expect(process.env.NEMOCLAW_ENDPOINT_URL).toBe(HOSTED_INFERENCE_ENDPOINT_URL);
         expect(process.env.NEMOCLAW_MODEL).toBe(HOSTED_INFERENCE_MODEL);
         expect(process.env.NEMOCLAW_COMPAT_MODEL).toBe(HOSTED_INFERENCE_MODEL);
+        expect(process.env.NEMOCLAW_PREFERRED_API).toBe("openai-completions");
         expect(process.env.COMPATIBLE_API_KEY).toBe("repo-hosted-key");
       },
     );
   });
 
+  it("does not override an explicit hosted inference API preference", () => {
+    withProviderEnv(
+      {
+        NVIDIA_INFERENCE_API_KEY: "repo-hosted-key",
+        NEMOCLAW_E2E_USE_HOSTED_INFERENCE: "1",
+        NEMOCLAW_PREFERRED_API: "openai-responses",
+      },
+      () => {
+        expect(stageHostedInferenceSourceSecretEnv()).toBe(true);
+        expect(process.env.NEMOCLAW_PREFERRED_API).toBe("openai-responses");
+      },
+    );
+  });
+
   it("keeps explicit cloud provider selection on the Build provider path", () => {
     withProviderEnv(
       {
diff --git a/src/lib/onboard/providers.ts b/src/lib/onboard/providers.ts
index 4ba749c2e1..18f835cb1b 100644
--- a/src/lib/onboard/providers.ts
+++ b/src/lib/onboard/providers.ts
@@ -254,6 +254,8 @@ function stageHostedInferenceSourceSecretEnv() {
     HOSTED_INFERENCE_MODEL;
   process.env.NEMOCLAW_MODEL = model;
   process.env.NEMOCLAW_COMPAT_MODEL = (process.env.NEMOCLAW_COMPAT_MODEL || "").trim() || model;
+  process.env.NEMOCLAW_PREFERRED_API =
+    (process.env.NEMOCLAW_PREFERRED_API || "").trim() || "openai-completions";
   process.env[HOSTED_INFERENCE_CREDENTIAL_ENV] = sourceKey;
   return true;
 }
diff --git a/test/e2e-advisor-dispatch.test.ts b/test/e2e-advisor-dispatch.test.ts
index a8b7e31bab..8859fc49c0 100644
--- a/test/e2e-advisor-dispatch.test.ts
+++ b/test/e2e-advisor-dispatch.test.ts
@@ -19,6 +19,23 @@ jobs:
   cloud-e2e:
     if: github.event_name != 'workflow_dispatch' || inputs.jobs == '' || contains(format(',{0},', inputs.jobs), ',cloud-e2e,')
     steps: []
+  cloud-onboard-e2e:
+    if: github.event_name != 'workflow_dispatch' || inputs.jobs == '' || contains(format(',{0},', inputs.jobs), ',cloud-onboard-e2e,')
+    uses: ./.github/workflows/e2e-script.yaml
+    with:
+      ref: \${{ inputs.target_ref || github.ref }}
+      nvidia_api_key: true
+    secrets: *nightly-e2e-default-secrets
+  launchable-smoke-e2e:
+    if: github.event_name != 'workflow_dispatch' || inputs.jobs == '' || contains(format(',{0},', inputs.jobs), ',launchable-smoke-e2e,')
+    runs-on: ubuntu-latest
+    steps:
+      - name: Run launchable install-flow smoke test
+        env:
+          NVIDIA_INFERENCE_API_KEY: \${{ (github.event_name != 'workflow_dispatch' || inputs.target_ref == '') && secrets.NVIDIA_INFERENCE_API_KEY || '' }}
+          NEMOCLAW_E2E_USE_HOSTED_INFERENCE: "1"
+          COMPATIBLE_API_KEY: \${{ (github.event_name != 'workflow_dispatch' || inputs.target_ref == '') && secrets.NVIDIA_INFERENCE_API_KEY || '' }}
+        run: bash test/e2e/test-launchable-smoke.sh
   report-to-pr:
     steps: []
   notify-on-failure:
@@ -68,6 +85,8 @@ describe("E2E advisor auto-dispatch planning", () => {
 
     expect(jobs).toContain("network-policy-e2e");
     expect(jobs).toContain("cloud-e2e");
+    expect(jobs).toContain("cloud-onboard-e2e");
+    expect(jobs).toContain("launchable-smoke-e2e");
     expect(jobs).not.toContain("report-to-pr");
     expect(jobs).not.toContain("notify-on-failure");
     expect(jobs).not.toContain("scorecard");
@@ -122,6 +141,56 @@ describe("E2E advisor auto-dispatch planning", () => {
     expect(plan.inputs?.jobs).toBe("network-policy-e2e,cloud-e2e");
   });
 
+  it("filters hosted-inference jobs that cannot receive secrets on target-ref dispatches", () => {
+    const workflowText = nightlyWorkflowText();
+    const plan = planAutoDispatch({
+      result: {
+        confidence: "high",
+        requiredTests: [
+          { id: "network-policy-e2e", workflow: "nightly-e2e.yaml" },
+          { id: "cloud-onboard-e2e", workflow: "nightly-e2e.yaml" },
+          { id: "launchable-smoke-e2e", workflow: "nightly-e2e.yaml" },
+        ],
+      },
+      workflowText,
+      event: pullRequest("MEMBER"),
+      env: {
+        GITHUB_EVENT_NAME: "pull_request",
+        GITHUB_REPOSITORY: "NVIDIA/NemoClaw",
+      },
+    });
+
+    expect(plan.status).toBe("ready");
+    expect(plan.jobs).toEqual(["network-policy-e2e"]);
+    expect(plan.inputs?.jobs).toBe("network-policy-e2e");
+    expect(plan.ignoredJobs).toEqual(["cloud-onboard-e2e", "launchable-smoke-e2e"]);
+    expect(plan.targetRefSecretBlockedJobs).toEqual(["cloud-onboard-e2e", "launchable-smoke-e2e"]);
+  });
+
+  it("skips cleanly when every recommended target-ref job requires withheld secrets", () => {
+    const workflowText = nightlyWorkflowText();
+    const plan = planAutoDispatch({
+      result: {
+        confidence: "high",
+        requiredTests: [
+          { id: "cloud-onboard-e2e", workflow: "nightly-e2e.yaml" },
+          { id: "launchable-smoke-e2e", workflow: "nightly-e2e.yaml" },
+        ],
+      },
+      workflowText,
+      event: pullRequest("MEMBER"),
+      env: {
+        GITHUB_EVENT_NAME: "pull_request",
+        GITHUB_REPOSITORY: "NVIDIA/NemoClaw",
+      },
+    });
+
+    expect(plan.status).toBe("skipped");
+    expect(plan.reason).toMatch(/hosted inference secrets are withheld/);
+    expect(plan.ignoredJobs).toEqual(["cloud-onboard-e2e", "launchable-smoke-e2e"]);
+    expect(plan.targetRefSecretBlockedJobs).toEqual(["cloud-onboard-e2e", "launchable-smoke-e2e"]);
+  });
+
   it("plans dispatch for allowlisted authors whose private org membership appears as contributor", () => {
     const workflowText = nightlyWorkflowText();
     const plan = planAutoDispatch({
diff --git a/test/e2e-script-workflow.test.ts b/test/e2e-script-workflow.test.ts
index 5fb007becb..5a0b151371 100644
--- a/test/e2e-script-workflow.test.ts
+++ b/test/e2e-script-workflow.test.ts
@@ -547,6 +547,7 @@ describe("E2E reusable workflow contract", () => {
     expect(runStep?.env?.NEMOCLAW_ENDPOINT_URL).toBe("https://inference-api.nvidia.com/v1");
     expect(runStep?.env?.NEMOCLAW_MODEL).toBe("nvidia/nvidia/nemotron-3-super-v3");
     expect(runStep?.env?.NEMOCLAW_COMPAT_MODEL).toBe("nvidia/nvidia/nemotron-3-super-v3");
+    expect(runStep?.env?.NEMOCLAW_PREFERRED_API).toBe("openai-completions");
     expect(runStep?.env?.COMPATIBLE_API_KEY).toBe(GUARDED_HOSTED_INFERENCE_SECRET);
     expect(runStep?.env?.GITHUB_TOKEN).toBeUndefined();
     expect(runStep?.env?.NEMOCLAW_RUN_E2E_SCENARIOS).toBe("1");
@@ -905,6 +906,7 @@ describe("E2E reusable workflow contract", () => {
     expect(exportStep?.run).toContain("NEMOCLAW_ENDPOINT_URL=https://inference-api.nvidia.com/v1");
     expect(exportStep?.run).toContain("NEMOCLAW_MODEL=nvidia/nvidia/nemotron-3-super-v3");
     expect(exportStep?.run).toContain("NEMOCLAW_COMPAT_MODEL=nvidia/nvidia/nemotron-3-super-v3");
+    expect(exportStep?.run).toContain("NEMOCLAW_PREFERRED_API=openai-completions");
     expect(exportStep?.run).toContain("COMPATIBLE_API_KEY=%s");
 
     expect(hostedJobs.length).toBeGreaterThan(20);
@@ -962,6 +964,7 @@ describe("E2E reusable workflow contract", () => {
       expect(step.env?.NEMOCLAW_ENDPOINT_URL, jobName).toBe("https://inference-api.nvidia.com/v1");
       expect(step.env?.NEMOCLAW_MODEL, jobName).toBe("nvidia/nvidia/nemotron-3-super-v3");
       expect(step.env?.NEMOCLAW_COMPAT_MODEL, jobName).toBe("nvidia/nvidia/nemotron-3-super-v3");
+      expect(step.env?.NEMOCLAW_PREFERRED_API, jobName).toBe("openai-completions");
       expect(step.env?.COMPATIBLE_API_KEY, jobName).toBe(GUARDED_HOSTED_INFERENCE_SECRET);
     }
 
diff --git a/test/e2e/lib/ci-compatible-inference.sh b/test/e2e/lib/ci-compatible-inference.sh
index 94c3cb1867..01b677d26c 100755
--- a/test/e2e/lib/ci-compatible-inference.sh
+++ b/test/e2e/lib/ci-compatible-inference.sh
@@ -37,6 +37,7 @@ nemoclaw_e2e_configure_compatible_inference() {
   export NEMOCLAW_ENDPOINT_URL="${NEMOCLAW_ENDPOINT_URL:-https://inference-api.nvidia.com/v1}"
   export NEMOCLAW_MODEL="${NEMOCLAW_MODEL:-${NEMOCLAW_CLOUD_EXPERIMENTAL_MODEL:-$NEMOCLAW_E2E_COMPATIBLE_INFERENCE_MODEL_DEFAULT}}"
   export NEMOCLAW_COMPAT_MODEL="${NEMOCLAW_COMPAT_MODEL:-$NEMOCLAW_MODEL}"
+  export NEMOCLAW_PREFERRED_API="${NEMOCLAW_PREFERRED_API:-openai-completions}"
   export COMPATIBLE_API_KEY="$NVIDIA_INFERENCE_API_KEY"
 }
 
diff --git a/tools/e2e-advisor/dispatch.mts b/tools/e2e-advisor/dispatch.mts
index b5637d5de8..44718097fd 100644
--- a/tools/e2e-advisor/dispatch.mts
+++ b/tools/e2e-advisor/dispatch.mts
@@ -60,6 +60,7 @@ type DispatchPlan = {
   jobs?: string[];
   ignoredJobs?: string[];
   recommendedJobs?: string[];
+  targetRefSecretBlockedJobs?: string[];
   dispatchableJobCount?: number;
   prNumber?: number;
   targetRef?: string;
@@ -260,16 +261,36 @@ export function planAutoDispatch({
     };
   }
 
-  const dispatchableJobs = extractDispatchableJobs(workflowText);
+  const targetRef = pr.head?.sha || pr.head?.ref || "";
+  const dispatchableJobInfos = extractDispatchableJobInfos(workflowText);
+  const dispatchableJobs = dispatchableJobInfos.map((job) => job.id);
   const recommendedJobs = collectRecommendedJobs(result, targetWorkflow);
-  const jobs = unique(recommendedJobs.filter((job) => dispatchableJobs.includes(job)));
-  const ignoredJobs = unique(recommendedJobs.filter((job) => !dispatchableJobs.includes(job)));
+  const dispatchableRecommendedJobs = unique(
+    recommendedJobs.filter((job) => dispatchableJobs.includes(job)),
+  );
+  const targetRefSecretBlockedJobs =
+    targetRef === ""
+      ? []
+      : dispatchableRecommendedJobs.filter((job) =>
+          dispatchableJobInfos.some((info) => info.id === job && info.targetRefSecretBlocked),
+        );
+  const jobs = unique(
+    dispatchableRecommendedJobs.filter((job) => !targetRefSecretBlockedJobs.includes(job)),
+  );
+  const ignoredJobs = unique([
+    ...recommendedJobs.filter((job) => !dispatchableJobs.includes(job)),
+    ...targetRefSecretBlockedJobs,
+  ]);
 
   if (jobs.length === 0) {
+    const onlyBlockedByTargetRefSecrets =
+      dispatchableRecommendedJobs.length > 0 &&
+      dispatchableRecommendedJobs.every((job) => targetRefSecretBlockedJobs.includes(job));
     return {
       ...base,
-      reason:
-        "no required advisor recommendations matched dispatchable jobs in the target workflow",
+      reason: onlyBlockedByTargetRefSecrets
+        ? "no required advisor recommendations can run with target_ref because hosted inference secrets are withheld"
+        : "no required advisor recommendations matched dispatchable jobs in the target workflow",
       prNumber: pr.number,
       authorAssociation,
       authorLogin,
@@ -277,10 +298,10 @@ export function planAutoDispatch({
       dispatchableJobCount: dispatchableJobs.length,
       recommendedJobs,
       ignoredJobs,
+      targetRefSecretBlockedJobs,
     };
   }
 
-  const targetRef = pr.head?.sha || pr.head?.ref || "";
   const dispatchRef = env.E2E_ADVISOR_AUTO_DISPATCH_REF || pr.base?.ref || DEFAULT_DISPATCH_REF;
   const advisorDispatchId = buildAdvisorDispatchId(pr.number, env);
   const inputs = {
@@ -299,6 +320,7 @@ export function planAutoDispatch({
     inputs,
     jobs,
     ignoredJobs,
+    targetRefSecretBlockedJobs,
     dispatchableJobCount: dispatchableJobs.length,
     prNumber: pr.number,
     targetRef,
@@ -311,11 +333,20 @@ export function planAutoDispatch({
 }
 
 export function extractDispatchableJobs(workflowText: string): string[] {
+  return extractDispatchableJobInfos(workflowText).map((job) => job.id);
+}
+
+type DispatchableJobInfo = {
+  id: string;
+  targetRefSecretBlocked: boolean;
+};
+
+function extractDispatchableJobInfos(workflowText: string): DispatchableJobInfo[] {
   const jobsBlockStart = workflowText.search(/^jobs:\s*$/m);
   if (jobsBlockStart === -1) return [];
 
   const lines = workflowText.slice(jobsBlockStart).split(/\r?\n/);
-  const jobs: string[] = [];
+  const jobs: DispatchableJobInfo[] = [];
   for (let index = 0; index < lines.length; index += 1) {
     const match = lines[index].match(/^  ([A-Za-z0-9_-]+):\s*$/);
     if (!match) continue;
@@ -328,10 +359,26 @@ export function extractDispatchableJobs(workflowText: string): string[] {
     }
     const body = bodyLines.join("\n");
     if (body.includes("inputs.jobs") && body.includes(`,${job},`)) {
-      jobs.push(job);
+      jobs.push({
+        id: job,
+        targetRefSecretBlocked: isTargetRefSecretBlockedJob(body),
+      });
     }
   }
-  return jobs.sort();
+  return jobs.sort((left, right) => left.id.localeCompare(right.id));
+}
+
+function isTargetRefSecretBlockedJob(body: string): boolean {
+  const callsScriptRunnerWithHostedInference =
+    body.includes("uses: ./.github/workflows/e2e-script.yaml") &&
+    /^\s+nvidia_api_key:\s*true\s*$/m.test(body);
+  if (callsScriptRunnerWithHostedInference) return true;
+
+  const hasTargetRefGuardedInferenceSecret =
+    body.includes("inputs.target_ref == ''") && body.includes("secrets.NVIDIA_INFERENCE_API_KEY");
+  const usesHostedInferenceSecret =
+    body.includes("NEMOCLAW_E2E_USE_HOSTED_INFERENCE") || body.includes("COMPATIBLE_API_KEY");
+  return hasTargetRefGuardedInferenceSecret && usesHostedInferenceSecret;
 }
 
 export function collectRecommendedJobs(
@@ -535,6 +582,14 @@ function renderDispatchSummary(result: DispatchPlan): string {
       `Ignored recommendations: ${result.ignoredJobs.map((job) => `\`${job}\``).join(", ")}`,
     );
   }
+  if (
+    Array.isArray(result.targetRefSecretBlockedJobs) &&
+    result.targetRefSecretBlockedJobs.length > 0
+  ) {
+    lines.push(
+      `Target-ref secret blocked jobs: ${result.targetRefSecretBlockedJobs.map((job) => `\`${job}\``).join(", ")}`,
+    );
+  }
   lines.push("");
   return `${lines.join("\n")}\n`;
 }