test: remove duplicate and theatrical tests (#2700)

- Remove 2-test "flag registration" block from custom-flag.test.ts — both
  assertions (KNOWN_FLAGS.has("--custom") and findUnknownFlag returning null)
  were already covered by the KNOWN_FLAGS completeness test in unknown-flags.test.ts.

- Fix stale KNOWN_FLAGS completeness test: it was testing only 18 of 26 known
  flags, making it always-pass when new flags are added to flags.ts without
  updating the test. Now the test is bidirectionally exhaustive — every flag in
  the expected list must be in KNOWN_FLAGS, and every flag in KNOWN_FLAGS must
  be in the expected list. This absorbs the --steps/--config coverage.

- Remove findUnknownFlag(["--steps"]) / findUnknownFlag(["--config"]) test from
  steps-flag.test.ts — now redundant since the exhaustive completeness test
  already exercises those flags.

Net: -3 tests removed, +18 expect() calls added (exhaustive bidirectional check).

Co-authored-by: spawn-qa-bot <qa@openrouter.ai>
Co-authored-by: L <6723574+louisgv@users.noreply.github.com>

packages/cli/src/__tests__/custom-flag.test.ts

@@ -1,23 +1,4 @@
 import { afterEach, describe, expect, it } from "bun:test";
-import { findUnknownFlag, KNOWN_FLAGS } from "../flags";
-
-describe("--custom flag", () => {
-  describe("flag registration", () => {
-    it("should be in KNOWN_FLAGS", () => {
-      expect(KNOWN_FLAGS.has("--custom")).toBe(true);
-    });
-
-    it("should not be detected as unknown flag", () => {
-      expect(
-        findUnknownFlag([
-          "claude",
-          "sprite",
-          "--custom",
-        ]),
-      ).toBeNull();
-    });
-  });
-});
 
 describe("AWS --custom prompts", () => {
   const savedCustom = process.env.SPAWN_CUSTOM;

packages/cli/src/__tests__/steps-flag.test.ts

@@ -1,22 +1,6 @@
 import { describe, expect, it } from "bun:test";
-import { findUnknownFlag } from "../flags";
 import { getAgentOptionalSteps, validateStepNames } from "../shared/agents";
 
-describe("--steps and --config flags", () => {
-  it("should recognize --steps and --config as known flags", () => {
-    expect(
-      findUnknownFlag([
-        "--steps",
-      ]),
-    ).toBeNull();
-    expect(
-      findUnknownFlag([
-        "--config",
-      ]),
-    ).toBeNull();
-  });
-});
-
 describe("validateStepNames", () => {
   it("should validate known steps for claude", () => {
     const { valid, invalid } = validateStepNames("claude", [

packages/cli/src/__tests__/unknown-flags.test.ts

@@ -189,6 +189,7 @@ describe("Unknown Flag Detection", () => {
 
 describe("KNOWN_FLAGS completeness", () => {
   it("should contain all expected flags", () => {
+    // This list must match flags.ts exactly — add here whenever KNOWN_FLAGS grows.
     const expected = [
       "--help",
       "-h",
@@ -213,12 +214,28 @@ describe("KNOWN_FLAGS completeness", () => {
       "--clear",
       "--custom",
       "--reauth",
+      "--zone",
+      "--region",
+      "--machine-type",
+      "--size",
       "--prune",
       "--json",
+      "--beta",
+      "--model",
+      "-m",
+      "--config",
+      "--steps",
+      "--user",
+      "-u",
     ];
+    // Every flag in the expected list must exist in KNOWN_FLAGS.
     for (const flag of expected) {
       expect(KNOWN_FLAGS.has(flag)).toBe(true);
     }
+    // Every flag in KNOWN_FLAGS must be in the expected list — catches silent additions.
+    for (const flag of KNOWN_FLAGS) {
+      expect(expected).toContain(flag);
+    }
   });
 });