fix(codex): carry the K12 short-window quota through cache and routing

package.json

@@ -1,6 +1,6 @@
 {
   "name": "@bitkyc08/opencodex",
-  "version": "2.24.2",
+  "version": "2.25.0",
   "description": "Universal provider proxy for OpenAI Codex & Claude Code — use any LLM with Codex CLI/App/SDK and Claude Code",
   "type": "module",
   "main": "./bin/package-main.mjs",

src/codex/quota.ts

@@ -246,6 +246,9 @@ export function setAccountQuotaFromParsed(
     if (existing?.monthlyPercent !== undefined) next.monthlyPercent = existing.monthlyPercent;
     if (existing?.monthlyResetAt !== undefined) next.monthlyResetAt = existing.monthlyResetAt;
     if (existing?.monthlyIsPrimaryWindow === true) next.monthlyIsPrimaryWindow = true;
+    if (existing?.shortPercent !== undefined) next.shortPercent = existing.shortPercent;
+    if (existing?.shortResetAt !== undefined) next.shortResetAt = existing.shortResetAt;
+    if (existing?.shortWindowSeconds !== undefined) next.shortWindowSeconds = existing.shortWindowSeconds;
     next.resetCredits = quota.resetCredits;
     accountQuota.set(accountId, next);
     schedulePersistAccountQuotas();
@@ -276,6 +279,15 @@ export function setAccountQuotaFromParsed(
     if (existing.monthlyIsPrimaryWindow === true) next.monthlyIsPrimaryWindow = true;
   }
 
+  // The burst (short) window is upstream-enforced on every plan, so its
+  // fields must survive parse → cache like the longer windows do: dropping
+  // them here reported shortPercent=null while the raw payload carried 0%,
+  // hiding imminent burst exhaustion from the API, dashboard, and routing
+  // (#2047).
+  if (quota.shortPercent !== undefined) next.shortPercent = quota.shortPercent;
+  if (quota.shortResetAt !== undefined) next.shortResetAt = quota.shortResetAt;
+  if (quota.shortWindowSeconds !== undefined) next.shortWindowSeconds = quota.shortWindowSeconds;
+
   if (quota.resetCredits !== undefined) next.resetCredits = quota.resetCredits;
   else if (existing?.resetCredits !== undefined) next.resetCredits = existing.resetCredits;
 

src/codex/routing.ts

@@ -322,14 +322,23 @@ function deleteScopedHealth(accountId: string, scope: CodexQuotaScope): void {
 export function computeCodexUsageScore(quota: {
   weeklyPercent?: number;
   monthlyPercent?: number;
+  shortPercent?: number;
 } | null, plan?: unknown): number {
   if (!quota) return CODEX_UNKNOWN_USAGE_SCORE;
+  // The burst window counts on every plan (upstream-enforced independently,
+  // see isCodexQuotaExhausted): a 0% long window with the burst window at
+  // 100% must not score as idle, or routing picks the account that 429s
+  // on the very next request (#2047).
+  const burst = typeof quota.shortPercent === "number" && Number.isFinite(quota.shortPercent)
+    ? quota.shortPercent
+    : undefined;
   if (isThirtyDayOnlyCodexPlan(plan)) {
-    return typeof quota.monthlyPercent === "number" && Number.isFinite(quota.monthlyPercent)
-      ? quota.monthlyPercent
-      : CODEX_UNKNOWN_USAGE_SCORE;
+    if (typeof quota.monthlyPercent !== "number" || !Number.isFinite(quota.monthlyPercent)) {
+      return burst !== undefined ? burst : CODEX_UNKNOWN_USAGE_SCORE;
+    }
+    return burst !== undefined ? Math.max(quota.monthlyPercent, burst) : quota.monthlyPercent;
   }
-  const values = [quota.weeklyPercent, quota.monthlyPercent]
+  const values = [quota.weeklyPercent, quota.monthlyPercent, burst]
     .filter((value): value is number => typeof value === "number" && Number.isFinite(value));
   return values.length > 0 ? Math.max(...values) : CODEX_UNKNOWN_USAGE_SCORE;
 }

tests/codex-routing.test.ts

@@ -122,6 +122,16 @@ describe("codex routing", () => {
     expect(computeCodexUsageScore({ weeklyPercent: 15 })).toBe(15);
   });
 
+  test("usage score counts the burst window on every plan (#2047)", () => {
+    // A long window at 0-1% with the burst window saturated must not score idle.
+    expect(computeCodexUsageScore({ weeklyPercent: 1, shortPercent: 100 })).toBe(100);
+    expect(computeCodexUsageScore({ monthlyPercent: 0, shortPercent: 87 }, "k12")).toBe(87);
+    expect(computeCodexUsageScore({ weeklyPercent: 15, shortPercent: 9 })).toBe(15);
+    // Missing burst data keeps the pre-existing behavior (long windows only).
+    expect(computeCodexUsageScore({ weeklyPercent: 15 })).toBe(15);
+    expect(computeCodexUsageScore(null)).toBe(CODEX_UNKNOWN_USAGE_SCORE);
+  });
+
   test("exact-account failures record health without rotating the active Pool account", () => {
     const transient = makeConfig({ upstreamFailoverThreshold: 1, activeCodexAccountId: "a" });
     const transientThread = "fixed-transient-thread";

tests/rate-limit-reset-credits.test.ts

@@ -445,3 +445,31 @@ describe("rate-limit reset credits", () => {
     });
   });
 });
+
+// ── short-window quota survives the account cache (#2047) ──────────────
+
+describe("short-window quota cache (#2047)", () => {
+  it("setAccountQuotaFromParsed carries the burst window into the cached snapshot", () => {
+    setAccountQuotaFromParsed("acct-short", {
+      weeklyPercent: 1,
+      weeklyResetAt: 1_000,
+      shortPercent: 0,
+      shortResetAt: 2_000,
+      shortWindowSeconds: 18_000,
+    });
+    const snap = getAccountQuota("acct-short");
+    expect(snap).not.toBeNull();
+    // 0% is data, not missing.
+    expect(snap!.shortPercent).toBe(0);
+    expect(snap!.shortResetAt).toBe(2_000);
+    expect(snap!.shortWindowSeconds).toBe(18_000);
+    expect(snap!.weeklyPercent).toBe(1);
+
+    // credits-only refresh must not drop the burst window either
+    setAccountQuotaFromParsed("acct-short", { resetCredits: 12 });
+    const carried = getAccountQuota("acct-short")!;
+    expect(carried.shortPercent).toBe(0);
+    expect(carried.shortWindowSeconds).toBe(18_000);
+    expect(carried.resetCredits).toBe(12);
+  });
+});