fix(macos): return false from isAppSetupCompleted for empty URL

desktop/macos/Desktop/Sources/APIClient.swift

@@ -3158,7 +3158,10 @@ extension APIClient {
 
   /// Checks if an external integration app's setup is complete
   func isAppSetupCompleted(url: String, uid: String) async -> Bool {
-    guard !url.isEmpty else { return true }
+    // An empty/unknown completion URL means setup cannot be verified, so report
+    // not-completed (consistent with the invalid-URL and network-failure paths
+    // below). Returning true here would wrongly mark an unconfigured app as set up.
+    guard !url.isEmpty else { return false }
     guard let fullUrl = URL(string: "\(url)?uid=\(uid)") else { return false }
     var request = URLRequest(url: fullUrl)
     request.httpMethod = "GET"

desktop/macos/Desktop/Tests/APIClientSetupCompletedTests.swift

@@ -0,0 +1,18 @@
+import XCTest
+
+@testable import Omi_Computer
+
+/// Regression coverage for `APIClient.isAppSetupCompleted(url:uid:)`.
+///
+/// An empty completion URL means setup cannot be verified, so the method must
+/// report `false` (not-completed) — matching its invalid-URL and
+/// network-failure paths. A previous `return true` would wrongly mark an
+/// unconfigured external integration app as already set up. The empty-URL case
+/// short-circuits before any network request, so this needs no HTTP mocking.
+final class APIClientSetupCompletedTests: XCTestCase {
+
+    func testEmptyURLReportsNotCompleted() async {
+        let completed = await APIClient.shared.isAppSetupCompleted(url: "", uid: "test-uid")
+        XCTAssertFalse(completed, "An empty completion URL must report setup as NOT completed")
+    }
+}