NVIDIA
diff --git a/‎docs/manage-sandboxes/uninstall-nemoclaw.mdx‎
Lines changed: 4 additions & 1 deletion b/‎docs/manage-sandboxes/uninstall-nemoclaw.mdx‎
Lines changed: 4 additions & 1 deletion
diff --git a/‎src/lib/actions/uninstall/portable-runtime-cleanup.test.ts‎
Lines changed: 94 additions & 0 deletions b/‎src/lib/actions/uninstall/portable-runtime-cleanup.test.ts‎
Lines changed: 94 additions & 0 deletions
diff --git a/‎src/lib/actions/uninstall/run-plan-gateway-service.test.ts‎
Lines changed: 98 additions & 83 deletions b/‎src/lib/actions/uninstall/run-plan-gateway-service.test.ts‎
Lines changed: 98 additions & 83 deletions
@@ -175,11 +175,13 @@ NemoClaw uses the pinned local Podman socket to remove only the exact receipt-ow
 Before every exact receipt-name OpenShell sandbox deletion, NemoClaw runs `openshell status -g <gateway>` to prove that the gateway name recorded by the matching registry row is connected.
 It then runs `openshell sandbox delete -g <gateway> <sandbox>` for that exact gateway and sandbox name.
 After deletion, bounded `openshell sandbox get -g <gateway> <sandbox>` probes must verify that the exact sandbox is absent.
+If the single delete command returns a nonzero status, NemoClaw still performs those read-only probes.
+Cleanup continues only when the connected recorded gateway returns the strict structured absence result for that exact sandbox; it never repeats the delete mutation.
 It never relies on or switches the globally selected gateway, and it never falls back to implicit deletion.
 A same-name sandbox on another gateway remains untouched.
 It never uses `openshell sandbox delete --all` during portable OpenShell cleanup.
 If explicit deletion reports an exact receipt sandbox as absent, cleanup continues only because the gateway-scoped status probe proved that exact gateway is connected and reachable.
-A missing gateway, missing provider, transport failure, verification timeout, generic or ambiguous `NotFound` result, or absence that the status probe cannot prove is fatal.
+A missing gateway, missing provider, transport failure, verification timeout, generic or ambiguous `NotFound` result during verification, or absence that the status probe cannot prove is fatal.
 Uninstall exits nonzero and preserves the portable registry row and container, selectors, receipts, `~/.config/nemoclaw/portable/containers.conf`, gateway state, shared OpenShell provider and gateway registrations, and unrelated OpenShell-only sandboxes.
 Portable cleanup retires only the exact receipt-owned portable resources.
 It preserves model stores and images, shared OpenShell provider and gateway registrations, the gateway service and process, shared OpenShell binaries and configuration, and unrelated OpenShell-registered sandboxes.
@@ -203,6 +205,7 @@ NemoClaw retains the record until a later completed onboarding durably supersede
 It contains a random local cleanup ID, receipt basenames derived from SHA-256 hashes of sandbox names, and length-framed, transaction-scoped, dictionary-testable pseudonymous fingerprints, but no raw sandbox or gateway names, absolute paths, environment values, configuration bytes, or secrets.
 A repeat invocation remains on portable cleanup and performs no Docker probe or mutation, implicit OpenShell cleanup, or model cleanup.
 If NemoClaw cannot prove the required portable authority or complete a cleanup step, uninstall exits nonzero.
+Before portable retirement completes, it also keeps the installed NemoClaw CLI and source so the same version-pinned command remains available for retry.
 It preserves any existing durable retirement record and any remaining receipts, `~/.config/nemoclaw/portable/containers.conf`, gateway state, and other retry evidence, even with `--destroy-user-data`.
 Resolve the reported authority or cleanup error, keep the preserved state intact, and rerun the same full uninstall command.
 This cleanup does not change Docker uninstall, externally supervised gateways, `--keep-openshell`, scoped sibling-gateway cleanup, or non-portable profiles.
 
@@ -329,6 +329,25 @@ function completeCleanup(input: PortableRuntimeCleanupInput, deps: PortableRunti
   return runPortableRuntimeCleanupTransaction(input, () => true, deps);
 }
 
+type PodmanObservation = ReturnType<NonNullable<PortableRuntimeCleanupDeps["podman"]>>;
+type PodmanObservationResponse = PodmanObservation | (() => PodmanObservation);
+
+function interceptSandboxDiscoveries(
+  test: ReturnType<typeof fixture>,
+  responses: readonly PodmanObservationResponse[],
+): () => number {
+  const originalPodman = test.deps.podman!;
+  let observations = 0;
+  test.deps.podman = vi.fn((args, env) => {
+    const command = args[0] === "--url" ? args.slice(2) : args;
+    const observesSandbox =
+      command[0] === "ps" && command.includes("label=openshell.ai/sandbox-name=alpha");
+    const response = observesSandbox ? responses[observations++] : undefined;
+    return (typeof response === "function" ? response() : response) ?? originalPodman(args, env);
+  });
+  return () => observations;
+}
+
 function addRetirementConfig(test: ReturnType<typeof fixture>): string {
   const target = path.join(test.homeDir, ".config/nemoclaw/portable/containers.conf");
   fs.mkdirSync(path.dirname(target), { mode: 0o700, recursive: true });
@@ -933,6 +952,81 @@ describe("portable runtime uninstall cleanup", () => {
     expect(fs.existsSync(`${test.registryFile}.lock`)).toBe(false);
   });
 
+  it("retries one failed read-only sandbox discovery under reasserted authority (#9499)", () => {
+    const test = fixture();
+    const observations = interceptSandboxDiscoveries(test, [
+      { status: 125, stdout: "", stderr: "transient local Podman observation failure" },
+    ]);
+
+    expect(completeCleanup(test.input, test.deps)).toEqual({
+      registryRemoved: true,
+      sandboxContainersRemoved: 1,
+      selectorsRemoved: ["CONTAINERS_CONF", "NETAVARK_FW"],
+    });
+    expect(observations()).toBe(4);
+    expect(test.containers.has(ALPHA_ID)).toBe(false);
+  });
+
+  it.each([
+    ["non-125 status", { status: 126, stdout: "", stderr: "permission denied" }],
+    ["spawn error", { status: null, stdout: "", stderr: "", error: new Error("spawn EACCES") }],
+  ] as const)("does not retry a %s sandbox discovery failure (#9499)", (_label, failure) => {
+    const test = fixture();
+    const observations = interceptSandboxDiscoveries(test, [failure]);
+
+    expect(() => completeCleanup(test.input, test.deps)).toThrow(
+      "Finding portable sandbox 'alpha' failed",
+    );
+    expect(observations()).toBe(1);
+    expect(test.containers.has(ALPHA_ID)).toBe(true);
+    expect(test.containers.has(REGISTRY_ID)).toBe(true);
+    expect(test.podmanCalls.some((args) => args[0] === "rm")).toBe(false);
+  });
+
+  it("stops a status-125 retry when socket authority changes before re-observation (#9499)", () => {
+    const test = fixture();
+    let authorityChanged = false;
+    const authorityError = "Podman socket authority changed after it was qualified";
+    const rejectChangedAuthority = () => {
+      throw new Error(authorityError);
+    };
+    const assertSocketAuthority = vi.fn(() =>
+      authorityChanged ? rejectChangedAuthority() : undefined,
+    );
+    test.deps.runtimeReadiness = { ...test.deps.runtimeReadiness, assertSocketAuthority };
+    const observations = interceptSandboxDiscoveries(test, [
+      () => {
+        authorityChanged = true;
+        return { status: 125, stdout: "", stderr: "transient local Podman observation failure" };
+      },
+    ]);
+
+    expect(() => completeCleanup(test.input, test.deps)).toThrow(authorityError);
+    expect(observations()).toBe(1);
+    expect(assertSocketAuthority).toHaveBeenCalled();
+    expect(test.containers.has(ALPHA_ID)).toBe(true);
+    expect(test.containers.has(REGISTRY_ID)).toBe(true);
+    expect(test.podmanCalls.some((args) => args[0] === "rm")).toBe(false);
+  });
+
+  it("stops without mutation after two failed read-only sandbox discoveries (#9499)", () => {
+    const test = fixture();
+    const failure = {
+      status: 125,
+      stdout: "",
+      stderr: "persistent local Podman observation failure",
+    };
+    const observations = interceptSandboxDiscoveries(test, [failure, failure]);
+
+    expect(() => completeCleanup(test.input, test.deps)).toThrow(
+      "Finding portable sandbox 'alpha' failed",
+    );
+    expect(observations()).toBe(2);
+    expect(test.containers.has(ALPHA_ID)).toBe(true);
+    expect(test.containers.has(REGISTRY_ID)).toBe(true);
+    expect(test.podmanCalls.some((args) => args[0] === "rm")).toBe(false);
+  });
+
   it("preserves changed current-user manager selector values (#9189)", () => {
     const test = fixture();
     test.selectors.set("CONTAINERS_CONF", "/home/test/user-containers.conf");
 
@@ -236,7 +236,7 @@ describe("uninstall OpenShell gateway user service", () => {
               ? ok(`${String(process.getuid?.() ?? -1)}\n`)
               : command === "ps" && args.includes("args=")
                 ? ok("/usr/local/bin/openshell-gateway --name nemoclaw --port 8080\n")
-              : ok(),
+                : ok(),
       },
       [{ name: "nemoclaw" }, { name: "sibling" }],
     );
@@ -335,7 +335,9 @@ describe("uninstall OpenShell gateway user service", () => {
             (command === "systemctl" &&
               args.includes("--property=MainPID") &&
               ok(`${String(pid)}\n`)) ||
-            (command === "ps" && args.includes("uid=") && ok(`${String(process.getuid?.() ?? -1)}\n`)) ||
+            (command === "ps" &&
+              args.includes("uid=") &&
+              ok(`${String(process.getuid?.() ?? -1)}\n`)) ||
             (command === "ps" &&
               args.includes("args=") &&
               ok("/usr/local/bin/openshell-gateway --name nemoclaw --port 8080\n")) ||
@@ -445,82 +447,82 @@ describe("uninstall OpenShell gateway user service", () => {
     { externallySupervised: false, keepOpenShell: false, mode: "managed cleanup" },
     { externallySupervised: false, keepOpenShell: true, mode: "--keep-openshell" },
     { externallySupervised: true, keepOpenShell: false, mode: "external supervision" },
-  ])("does not mutate scoped resources with an unproven sandbox namespace under $mode (#8663)", ({
-    externallySupervised,
-    keepOpenShell,
-  }) => {
-    const test = fixture(true);
-    const servicePath = writeManagedService(test);
-    const configPath = writeGatewayState(test);
-    const registryPath = writeSelectedSandboxRegistry(test, "my-assistant");
-    const registryBefore = fs.readFileSync(registryPath, "utf-8");
-    fs.writeFileSync(configPath, '[openshell.drivers.docker]\nsandbox_namespace = "default"\n');
-    const calls: string[][] = [];
-    const errors: string[] = [];
-
-    const externalAuthority = externallySupervised
-      ? {
-          resolveGatewayTeardownAuthority: ({
-            gatewayName,
-            gatewayPort,
-          }: {
-            gatewayName: string;
-            gatewayPort: number;
-          }) => ({
-            gatewayName,
-            gatewayPort,
-            mode: "externally-supervised" as const,
-            source: "declared" as const,
-            endpoint: `http://127.0.0.1:${String(gatewayPort)}`,
-            stateDir: path.dirname(configPath),
-            supervisor: {
-              kind: "systemd-user" as const,
-              serviceName: "external-openshell.service",
-              execPath: "/usr/local/bin/openshell-gateway",
-            },
-            requiredCapabilities: [],
-          }),
-        }
-      : {};
-    const deps: Partial<UninstallRunDeps> = {
-      commandExists: (command) => command === "systemctl",
-      run: (command, args) => {
-        calls.push([command, ...args]);
-        return ok();
-      },
-      error: (message) => errors.push(message),
-      ...externalAuthority,
-    };
-
-    const result = uninstall(test, keepOpenShell, deps, [
-      { name: "nemoclaw" },
-      { name: "nemoclaw-8081" },
-    ]);
-
-    expect(result.exitCode).toBe(1);
-    expect(fs.existsSync(servicePath)).toBe(true);
-    expect(fs.existsSync(configPath)).toBe(true);
-    expect(fs.readFileSync(registryPath, "utf-8")).toBe(registryBefore);
-    expect(
-      calls.some(([command, resource]) => command === "openshell" && resource === "sandbox"),
-    ).toBe(false);
-    expect(
-      calls.some(
-        ([command, resource, action]) =>
-          command === "openshell" && resource === "gateway" && action === "remove",
-      ),
-    ).toBe(false);
-    expect(
-      calls.some(
-        ([command, ...args]) => command === "systemctl" && !args.includes("--property=MainPID"),
-      ),
-    ).toBe(false);
-    expect(errors).toContain(
-      externallySupervised
-        ? "Refusing scoped gateway cleanup because the externally supervised process's loaded sandbox namespace cannot be proven."
-        : "Refusing scoped gateway cleanup because its sandbox namespace cannot be proven.",
-    );
-  });
+  ])(
+    "does not mutate scoped resources with an unproven sandbox namespace under $mode (#8663)",
+    ({ externallySupervised, keepOpenShell }) => {
+      const test = fixture(true);
+      const servicePath = writeManagedService(test);
+      const configPath = writeGatewayState(test);
+      const registryPath = writeSelectedSandboxRegistry(test, "my-assistant");
+      const registryBefore = fs.readFileSync(registryPath, "utf-8");
+      fs.writeFileSync(configPath, '[openshell.drivers.docker]\nsandbox_namespace = "default"\n');
+      const calls: string[][] = [];
+      const errors: string[] = [];
+
+      const externalAuthority = externallySupervised
+        ? {
+            resolveGatewayTeardownAuthority: ({
+              gatewayName,
+              gatewayPort,
+            }: {
+              gatewayName: string;
+              gatewayPort: number;
+            }) => ({
+              gatewayName,
+              gatewayPort,
+              mode: "externally-supervised" as const,
+              source: "declared" as const,
+              endpoint: `http://127.0.0.1:${String(gatewayPort)}`,
+              stateDir: path.dirname(configPath),
+              supervisor: {
+                kind: "systemd-user" as const,
+                serviceName: "external-openshell.service",
+                execPath: "/usr/local/bin/openshell-gateway",
+              },
+              requiredCapabilities: [],
+            }),
+          }
+        : {};
+      const deps: Partial<UninstallRunDeps> = {
+        commandExists: (command) => command === "systemctl",
+        run: (command, args) => {
+          calls.push([command, ...args]);
+          return ok();
+        },
+        error: (message) => errors.push(message),
+        ...externalAuthority,
+      };
+
+      const result = uninstall(test, keepOpenShell, deps, [
+        { name: "nemoclaw" },
+        { name: "nemoclaw-8081" },
+      ]);
+
+      expect(result.exitCode).toBe(1);
+      expect(fs.existsSync(servicePath)).toBe(true);
+      expect(fs.existsSync(configPath)).toBe(true);
+      expect(fs.readFileSync(registryPath, "utf-8")).toBe(registryBefore);
+      expect(
+        calls.some(([command, resource]) => command === "openshell" && resource === "sandbox"),
+      ).toBe(false);
+      expect(
+        calls.some(
+          ([command, resource, action]) =>
+            command === "openshell" && resource === "gateway" && action === "remove",
+        ),
+      ).toBe(false);
+      expect(
+        calls.some(
+          ([command, ...args]) => command === "systemctl" && !args.includes("--property=MainPID"),
+        ),
+      ).toBe(false);
+      expect(errors).toContain(
+        externallySupervised
+          ? "Refusing scoped gateway cleanup because the externally supervised process's loaded sandbox namespace cannot be proven."
+          : "Refusing scoped gateway cleanup because its sandbox namespace cannot be proven.",
+      );
+    },
+  );
 
   it("preserves the marked Linux unit when scoped sandbox deletion fails (#8220)", () => {
     const test = fixture(true);
@@ -698,18 +700,31 @@ describe("uninstall OpenShell gateway user service", () => {
     const test = fixture();
     const servicePath = writeManagedService(test);
     const errors: string[] = [];
+    const run = vi.fn((command: string, args: string[]) =>
+      command === "systemctl" && args.includes("disable")
+        ? { status: 1, stdout: "", stderr: "failed" }
+        : ok(),
+    );
 
     const result = uninstall(test, false, {
-      commandExists: (command) => command === "systemctl",
+      commandExists: (command) => ["systemctl", "npm"].includes(command),
       error: (line) => errors.push(line),
-      run: (command, args) =>
-        command === "systemctl" && args.includes("disable")
-          ? { status: 1, stdout: "", stderr: "failed" }
-          : ok(),
+      run,
     });
 
     expect(result.exitCode).toBe(1);
     expect(fs.existsSync(servicePath)).toBe(true);
+    const failedDisableIndex = run.mock.calls.findIndex(
+      ([command, args]) => command === "systemctl" && args.includes("disable"),
+    );
+    const npmCleanupIndex = run.mock.calls.findIndex(
+      ([command, args], index) =>
+        index > failedDisableIndex &&
+        command === "npm" &&
+        args.join(" ") === "uninstall -g --loglevel=error nemoclaw",
+    );
+    expect(failedDisableIndex).toBeGreaterThanOrEqual(0);
+    expect(npmCleanupIndex).toBeGreaterThan(failedDisableIndex);
     expect(errors).toContain(
       "Uninstall completed with errors. Some state may remain on disk; see warnings above.",
     );