nemoclaw/src/blueprint/runner.test.ts (1)

577-641: ⚠️ Potential issue | 🟠 Major

Please restore regression coverage for apply --plan rejection.

main no longer tests the unsupported --plan path, but runtime still rejects it in actionApply. This leaves CLI parse/dispatch behavior unguarded.

Proposed test addition

   describe("main (CLI)", () => {
@@
     it("parses apply with --profile and --endpoint-url", async () => {
       await main(["apply", "--profile", "default", "--endpoint-url", "https://override.test/v1"]);
       expect(mockedValidateEndpoint).toHaveBeenCalledWith("https://override.test/v1");
       expect(stdoutText()).toContain("PROGRESS:100:Apply complete");
     });
+
+    it("rejects apply when --plan is provided (not yet implemented)", async () => {
+      await expect(
+        main(["apply", "--profile", "default", "--plan", "/tmp/plan.json"]),
+      ).rejects.toThrow(/--plan is not yet implemented/);
+    });

🤖 Prompt for AI Agents

Verify each finding against the current code and only fix it if needed.

In `@nemoclaw/src/blueprint/runner.test.ts` around lines 577 - 641, Add a test in
the existing "main (CLI)" suite that exercises the unsupported apply --plan
path: call main with arguments like ["apply","--plan","some-plan.json"] (after
the existing beforeEach setup) and assert it rejects with an error containing
"--plan" (or the exact rejection text emitted by actionApply); this restores
regression coverage for the main -> actionApply dispatch path and ensures CLI
parsing still rejects the --plan option at runtime.

test/cli.test.js (1)

16-27: ⚠️ Potential issue | 🔴 Critical

Critical bug: spawnSync is misconfigured — tests will fail with TypeError: r.out.includes is not a function.

The current implementation has multiple issues:

1. spawnSync does not throw exceptions — unlike execSync, it always returns a result object with an error property. The try-catch block will never catch non-zero exits; the returned out is always the full result object {error, status, stdout, stderr, ...}, which has no .includes() method.

2. Tests call .includes() on an object — every test assertion like r.out.includes("Getting Started") will fail at runtime with TypeError: r.out.includes is not a function.

3. Missing shell: true — without it, spawnSync treats the string as a literal executable name (looking for a file named node "${CLI}" ${args}), resulting in ENOENT instead of executing the shell command.

To fix, use:

Corrected implementation

 function runWithEnv(args, env = {}, timeout = 10000) {
-  try {
-    const out = spawnSync(`node "${CLI}" ${args}`, {
-      encoding: "utf-8",
-      timeout,
-      env: { ...process.env, HOME: "/tmp/nemoclaw-cli-test-" + Date.now(), ...env },
-    });
-    return { code: 0, out };
-  } catch (err) {
-    return { code: err.status, out: (err.stdout || "") + (err.stderr || "") };
-  }
+  const result = spawnSync(`node "${CLI}" ${args}`, {
+    shell: true,
+    encoding: "utf-8",
+    timeout,
+    env: { ...process.env, HOME: "/tmp/nemoclaw-cli-test-" + Date.now(), ...env },
+  });
+  const out = (result.stdout || "") + (result.stderr || "");
+  return { code: result.status ?? 1, out };
 }

🤖 Prompt for AI Agents

Verify each finding against the current code and only fix it if needed.

In `@test/cli.test.js` around lines 16 - 27, The runWithEnv function misuses
spawnSync: it never throws, returns a result object (so tests calling
r.out.includes fail), and the command string needs shell: true; fix runWithEnv
by calling spawnSync with shell: true (or pass command and args as an array),
then read the returned result.stdout/stderr (convert to string) and
result.status/result.error to determine exit code; return { code: <numeric
status or error.status>, out: <stdout + stderr as string> } so callers can
safely call r.out.includes; update references in runWithEnv to use the result
object fields instead of assuming spawnSync throws.