fix(usage): reuse Gemini CLI project ID for quota checks (#4454)

Integrated into release/v3.8.32

CHANGELOG.md

@@ -16,6 +16,7 @@ _In development — bullets added per PR; finalized at release._
 
 - **fix(embeddings):** forward output dimensions to Gemini for consistent embedding dims. (thanks @nguyenha935)
 - **fix(translator):** sanitize Read tool args from non-Anthropic models to prevent retry loops. (thanks @GodrezJr2)
+- **fix(usage):** reuse Gemini CLI project ID for quota checks (avoid re-discovery). (thanks @Delcado19)
 - **fix(combo): round-robin members fail over faster under concurrency saturation via a configurable queue depth** — when a round-robin combo member was saturated, requests sat in the per-model semaphore's **unbounded** queue and only failed over to the next member after the full `queueTimeoutMs` (default 30s) elapsed — so a burst of agentic requests deep-queued one hot member instead of spilling to healthy ones. The per-model semaphore now accepts a bounded queue depth and emits `SEMAPHORE_QUEUE_FULL` once it is full (the round-robin loop already cascades on that code), so a configured low depth fails over immediately. A new `queueDepth` combo-config knob (global default / provider override / per-combo, default **20** for backward compatibility; **0** = never queue → fail over now) is exposed in Settings → Combo Defaults. ([#3872](https://github.com/diegosouzapw/OmniRoute/issues/3872) — thanks @KooshaPari)
 
 ---

open-sse/services/usage.ts

@@ -1812,6 +1812,23 @@ function inferGitHubPlanName(data: JsonRecord, premiumQuota: UsageQuota | null):
 const _geminiCliSubCache = new Map<string, SubscriptionCacheEntry>();
 const GEMINI_CLI_CACHE_TTL_MS = 5 * 60 * 1000; // 5 minutes
 
+/**
+ * Normalize a Cloud Code project value into a trimmed string (or null).
+ * The upstream `loadCodeAssist` endpoint returns the project either as a bare
+ * string or as an object of the form `{ id: "..." }`, and stored connection
+ * project ids can carry stray whitespace. Centralized here so the Gemini CLI
+ * usage path matches the executor/oauth normalization already shipped in
+ * `open-sse/executors/gemini-cli.ts` and `src/lib/oauth/services/gemini.ts`.
+ */
+function normalizeCloudCodeProjectId(project: unknown): string | null {
+  if (typeof project === "string") return project.trim() || null;
+  if (project && typeof project === "object") {
+    const candidate = (project as { id?: unknown }).id;
+    if (typeof candidate === "string") return candidate.trim() || null;
+  }
+  return null;
+}
+
 /**
  * Gemini CLI Usage — fetch per-model quota from Cloud Code Assist API.
  * Gemini CLI and Antigravity share the same upstream (cloudcode-pa.googleapis.com),
@@ -1827,17 +1844,28 @@ async function getGeminiUsage(
   }
 
   try {
-    const subscriptionInfo = await getGeminiCliSubscriptionInfoCached(accessToken);
-    const projectId =
-      connectionProjectId ||
-      providerSpecificData?.projectId ||
-      toRecord(subscriptionInfo).cloudaicompanionProject ||
-      null;
-
-    const plan = getGeminiCliPlanLabel(subscriptionInfo);
+    // #1271: the OAuth save path stores `projectId` on the connection (not always in
+    // `providerSpecificData`), and `loadCodeAssist` may return the project either as a
+    // bare string or wrapped in `{ id: "..." }`. Normalize both so the quota lookup
+    // reuses the stored project id and skips a redundant `loadCodeAssist` round-trip
+    // when it is already known.
+    let projectId =
+      normalizeCloudCodeProjectId(connectionProjectId) ||
+      normalizeCloudCodeProjectId(providerSpecificData?.projectId);
+    let plan = "Free";
 
     if (!projectId) {
-      return { plan, message: "Gemini CLI project ID not available." };
+      const subscriptionInfo = await getGeminiCliSubscriptionInfoCached(accessToken);
+      projectId = normalizeCloudCodeProjectId(toRecord(subscriptionInfo).cloudaicompanionProject);
+      plan = getGeminiCliPlanLabel(subscriptionInfo);
+    }
+
+    if (!projectId) {
+      return {
+        plan,
+        message:
+          "Gemini CLI project ID not available. Reconnect Gemini CLI, or configure a Google Cloud project with Gemini Code Assist access before checking quota.",
+      };
     }
 
     // Use retrieveUserQuota (same endpoint as Gemini CLI /stats command).

tests/unit/gemini-usage-projectid.test.ts

@@ -0,0 +1,145 @@
+/**
+ * Ported from decolua/9router#1428 by @Delcado19 — reuses stored Gemini CLI
+ * project IDs for quota checks and normalizes {id: ...} object shapes that
+ * loadCodeAssist returns.
+ */
+import test from "node:test";
+import assert from "node:assert/strict";
+
+const usage = await import("../../open-sse/services/usage.ts");
+
+function jsonResponse(body: unknown, status = 200): Response {
+  return new Response(JSON.stringify(body), {
+    status,
+    headers: { "content-type": "application/json" },
+  });
+}
+
+test("getUsageForProvider(gemini-cli) reuses the projectId stored on the connection", async () => {
+  const originalFetch = globalThis.fetch;
+  const calls: { url: string; init?: RequestInit }[] = [];
+  globalThis.fetch = (async (input: RequestInfo | URL, init?: RequestInit) => {
+    const url = typeof input === "string" ? input : input.toString();
+    calls.push({ url, init });
+    return jsonResponse({
+      buckets: [
+        {
+          modelId: "gemini-3-flash-preview",
+          remainingFraction: 0.75,
+          resetTime: "2026-05-25T12:00:00Z",
+        },
+      ],
+    });
+  }) as typeof fetch;
+
+  try {
+    const result = (await usage.getUsageForProvider({
+      id: "gemini-cli-stored",
+      provider: "gemini-cli",
+      accessToken: "token",
+      projectId: "cloud-code-project",
+    })) as { quotas?: Record<string, { remainingPercentage?: number }> };
+
+    // Only the retrieveUserQuota call — no loadCodeAssist round-trip,
+    // because the stored projectId short-circuits it.
+    assert.equal(calls.length, 1, "must skip loadCodeAssist when projectId is stored");
+    assert.equal(
+      calls[0].url,
+      "https://cloudcode-pa.googleapis.com/v1internal:retrieveUserQuota"
+    );
+    assert.equal(
+      JSON.parse(String(calls[0].init?.body)).project,
+      "cloud-code-project",
+      "must pass the stored projectId into retrieveUserQuota"
+    );
+    assert.equal(result.quotas?.["gemini-3-flash-preview"]?.remainingPercentage, 75);
+  } finally {
+    globalThis.fetch = originalFetch;
+  }
+});
+
+test("getUsageForProvider(gemini-cli) normalizes project objects returned by loadCodeAssist", async () => {
+  const originalFetch = globalThis.fetch;
+  const calls: { url: string; init?: RequestInit }[] = [];
+  globalThis.fetch = (async (input: RequestInfo | URL, init?: RequestInit) => {
+    const url = typeof input === "string" ? input : input.toString();
+    calls.push({ url, init });
+    if (url.endsWith("loadCodeAssist")) {
+      return jsonResponse({
+        cloudaicompanionProject: { id: "project-from-load" },
+        currentTier: { name: "Free" },
+      });
+    }
+    return jsonResponse({ buckets: [] });
+  }) as typeof fetch;
+
+  try {
+    await usage.getUsageForProvider({
+      id: "gemini-cli-obj-shape",
+      provider: "gemini-cli",
+      accessToken: "token-obj-shape",
+    });
+
+    const quotaCall = calls.find((c) => c.url.endsWith("retrieveUserQuota"));
+    assert.ok(quotaCall, "quota lookup must occur after loadCodeAssist resolves");
+    assert.equal(
+      JSON.parse(String(quotaCall!.init?.body)).project,
+      "project-from-load",
+      "must unwrap {id: ...} into the bare project id"
+    );
+  } finally {
+    globalThis.fetch = originalFetch;
+  }
+});
+
+test("getUsageForProvider(gemini-cli) trims whitespace-padded stored project ids", async () => {
+  const originalFetch = globalThis.fetch;
+  const calls: { url: string; init?: RequestInit }[] = [];
+  globalThis.fetch = (async (input: RequestInfo | URL, init?: RequestInit) => {
+    const url = typeof input === "string" ? input : input.toString();
+    calls.push({ url, init });
+    return jsonResponse({ buckets: [] });
+  }) as typeof fetch;
+
+  try {
+    await usage.getUsageForProvider({
+      id: "gemini-cli-padded",
+      provider: "gemini-cli",
+      accessToken: "token-padded",
+      projectId: "  padded-project  ",
+    });
+
+    const quotaCall = calls.find((c) => c.url.endsWith("retrieveUserQuota"));
+    assert.ok(quotaCall, "quota lookup must run with the trimmed project id");
+    assert.equal(JSON.parse(String(quotaCall!.init?.body)).project, "padded-project");
+  } finally {
+    globalThis.fetch = originalFetch;
+  }
+});
+
+test("getUsageForProvider(gemini-cli) returns actionable guidance when no project id is available", async () => {
+  const originalFetch = globalThis.fetch;
+  globalThis.fetch = (async () => jsonResponse({})) as typeof fetch;
+
+  try {
+    const result = (await usage.getUsageForProvider({
+      id: "gemini-cli-no-project",
+      provider: "gemini-cli",
+      accessToken: "token-no-project",
+    })) as { message?: string };
+
+    assert.ok(result.message, "must surface a message when no projectId is resolvable");
+    assert.match(
+      result.message!,
+      /Reconnect Gemini CLI/i,
+      "error must guide the operator to reconnect"
+    );
+    assert.match(
+      result.message!,
+      /Gemini Code Assist/i,
+      "error must mention the Code Assist project requirement"
+    );
+  } finally {
+    globalThis.fetch = originalFetch;
+  }
+});