fix(test): stop the Alibaba allowlist test from expiring with the catalog (#11867)

`Unit Tests (1/8)` went red on 2026-08-28 across every PR and on main, with
nothing changed — the clock had moved past the shipped catalog's expiry:

  config/alibaba-free-tier-allowlist.json → "validUntil": "2026-08-27"

isAlibabaFreeTierAllowlistPackValid() compares that against Date.now(), so from
28/08 loadAlibabaFreeTierAllowlistPack() returns null and the old
assert.ok(pack) could never pass again. Refreshing the date would only reschedule
the same break.

Production was never affected: resolveActiveAllowlistPack() falls back to the
embedded list when a pack expires, which is the intended design. The defect was
the test asserting the shipped catalog is currently fresh — a data property, not
a behavioral contract.

The test now writes its own packs to a temp dir with dates it controls, and
pins both halves of the contract:

  - inside the validity window, the pack REPLACES the embedded list (anchored on
    a model that exists nowhere else, so loading alone cannot satisfy it);
  - once expired, the pack is ignored and the embedded list serves.

That second path is what production has been running since 27/08 and had no
coverage at all, which is why the expiry surfaced as a red test rather than as
understood behavior. A third case pins the comparison against an injected
instant, including the no-expiry pack that never goes stale.

Whether the curated free-tier catalog still matches reality — and so deserves a
freshly dated pack — is a data question left to the operator in #11866.

Closes #11866

changelog.d/fixes/11866-alibaba-allowlist-test-timebomb.md

@@ -0,0 +1,5 @@
+- Fixed the Alibaba free-tier allowlist test that went red on its own once the
+  shipped catalog's `validUntil` (2026-08-27) passed, leaving every PR and `main`
+  with a failing `Unit Tests (1/8)`. The test now builds its own packs with dates
+  it controls, and covers the expired-pack fallback that production has actually
+  been serving.

tests/unit/alibaba-free-tier-allowlist.test.ts

@@ -7,6 +7,9 @@
  */
 import { test } from "node:test";
 import assert from "node:assert/strict";
+import { mkdtempSync, rmSync, writeFileSync } from "node:fs";
+import { tmpdir } from "node:os";
+import { join } from "node:path";
 import {
   ALIBABA_FREE_TIER_TEXT_CAPABLE_MODELS,
   ALIBABA_NO_FREE_TIER_TEXT_MODELS,
@@ -27,18 +30,90 @@ test("built-in allowlist includes operator free models and excludes paid blockli
   assert.equal(isAlibabaBuiltinFreeTierTextModel("qwen3.7-max"), false);
 });
 
-test("allowlist JSON pack overrides embedded lists when valid", () => {
+/**
+ * The shipped `config/alibaba-free-tier-allowlist.json` carries a `validUntil`,
+ * so asserting against it made this test a time bomb: it went red on its own on
+ * 2026-08-28, the day after the pack expired, and stayed red on every PR and on
+ * `main` (#11866). Nothing had changed — the clock moved.
+ *
+ * Production was never affected: an expired pack falls back to the embedded
+ * list by design. So the contract worth pinning is the BEHAVIOR on both sides of
+ * the expiry, with packs this test owns and dates it controls — never the
+ * freshness of the catalog that ships in the repo.
+ */
+function withAllowlistPack(
+  pack: Record<string, unknown>,
+  assertions: () => void
+): void {
+  const dir = mkdtempSync(join(tmpdir(), "alibaba-allowlist-"));
+  const packPath = join(dir, "allowlist.json");
+  writeFileSync(packPath, JSON.stringify(pack), "utf8");
+
   const previousPath = process.env.ALIBABA_FREE_TIER_ALLOWLIST_PATH;
-  const packPath = `${process.cwd()}/config/alibaba-free-tier-allowlist.json`;
   process.env.ALIBABA_FREE_TIER_ALLOWLIST_PATH = packPath;
   resetAlibabaFreeTierAllowlistCache();
+  try {
+    assertions();
+  } finally {
+    if (previousPath) process.env.ALIBABA_FREE_TIER_ALLOWLIST_PATH = previousPath;
+    else delete process.env.ALIBABA_FREE_TIER_ALLOWLIST_PATH;
+    resetAlibabaFreeTierAllowlistCache();
+    rmSync(dir, { recursive: true, force: true });
+  }
+}
 
-  const pack = loadAlibabaFreeTierAllowlistPack();
-  assert.ok(pack);
-  assert.ok(isAlibabaFreeTierAllowlistPackValid(pack!));
-  assert.ok(pack!.capable.includes("qwen3.6-plus"));
+test("allowlist JSON pack overrides embedded lists while it is still valid", () => {
+  withAllowlistPack(
+    {
+      asOf: "2026-07-28",
+      validUntil: "2999-01-01",
+      capable: ["pack-only-capable-model", "qwen3.6-plus"],
+      noFreeTier: ["pack-only-paid-model"],
+    },
+    () => {
+      const pack = loadAlibabaFreeTierAllowlistPack();
+      assert.ok(pack, "a pack inside its validity window must load");
+      assert.ok(isAlibabaFreeTierAllowlistPackValid(pack!));
+      assert.ok(pack!.capable.includes("qwen3.6-plus"));
+      // Positive anchor: the pack must actually REPLACE the embedded list, not
+      // merely load. `pack-only-capable-model` exists nowhere else.
+      assert.equal(isAlibabaBuiltinFreeTierTextModel("pack-only-capable-model"), true);
+      assert.equal(isAlibabaBuiltinNoFreeTierTextModel("pack-only-paid-model"), true);
+    }
+  );
+});
 
-  if (previousPath) process.env.ALIBABA_FREE_TIER_ALLOWLIST_PATH = previousPath;
-  else delete process.env.ALIBABA_FREE_TIER_ALLOWLIST_PATH;
-  resetAlibabaFreeTierAllowlistCache();
+test("an expired allowlist pack is ignored and the embedded list serves instead", () => {
+  // This is the path production has actually been on since 2026-08-27, and it
+  // had no coverage at all — which is why the expiry surfaced as a red test
+  // rather than as a deliberate, understood fallback.
+  withAllowlistPack(
+    {
+      asOf: "2026-07-28",
+      validUntil: "2026-08-27",
+      capable: ["pack-only-capable-model"],
+      noFreeTier: ["pack-only-paid-model"],
+    },
+    () => {
+      assert.equal(loadAlibabaFreeTierAllowlistPack(), null, "expired pack must not load");
+      assert.equal(isAlibabaBuiltinFreeTierTextModel("pack-only-capable-model"), false);
+      // The embedded list must be what answers once the pack is rejected.
+      assert.equal(isAlibabaBuiltinFreeTierTextModel("qwen3.6-plus"), true);
+      assert.equal(isAlibabaBuiltinNoFreeTierTextModel("qwen3.7-max"), true);
+    }
+  );
+});
+
+test("isAlibabaFreeTierAllowlistPackValid compares against the instant it is given", () => {
+  const pack = { asOf: "2026-07-28", validUntil: "2026-08-27", capable: ["x"], noFreeTier: [] };
+  assert.equal(isAlibabaFreeTierAllowlistPackValid(pack, Date.parse("2026-08-26")), true);
+  assert.equal(isAlibabaFreeTierAllowlistPackValid(pack, Date.parse("2026-08-28")), false);
+  // No expiry declared means the pack never goes stale on its own.
+  assert.equal(
+    isAlibabaFreeTierAllowlistPackValid(
+      { asOf: "2026-07-28", capable: ["x"], noFreeTier: [] },
+      Date.parse("2999-01-01")
+    ),
+    true
+  );
 });