fix(codex): route bare codex-auto-review to the Codex provider (#4135)

open-sse/providers/registry/codex.js

@@ -65,6 +65,9 @@ export default {
     { id: "gpt-5.4-mini-review", name: "GPT 5.4 Mini Review", upstreamModelId: "gpt-5.4-mini", quotaFamily: "review" },
     { id: "gpt-5.3-codex-spark", name: "GPT 5.3 Codex Spark" },
     { id: "gpt-5.3-codex-spark-review", name: "GPT 5.3 Codex Spark Review", upstreamModelId: "gpt-5.3-codex-spark", quotaFamily: "review" },
+    // Codex CLI's auto-review virtual model. Unlike the "-review" variants above it is not derived
+    // from a base model, so it is forwarded verbatim instead of having "-review" stripped (#1398).
+    { id: "codex-auto-review", name: "Codex Auto Review", upstreamModelId: "codex-auto-review", quotaFamily: "review" },
     { id: "gpt-image-2.5", name: "GPT Image 2.5", capabilities: ["text2img","edit","multiImage"], params: ["size","quality","background","image_detail","output_format"], kind: "image" },
     { id: "gpt-image-2.5-flare", name: "GPT Image 2.5 Flare", capabilities: ["text2img","edit","multiImage"], params: ["size","quality","background","image_detail","output_format"], kind: "image" },
     { id: "gpt-image-2.5-sunburst", name: "GPT Image 2.5 Sunburst", capabilities: ["text2img","edit","multiImage"], params: ["size","quality","background","image_detail","output_format"], kind: "image" },

open-sse/services/model.js

@@ -124,6 +124,8 @@ export async function getModelInfoCore(modelStr, aliasesOrGetter) {
 
 // Config-driven prefix → provider inference (first match wins, fallback "openai").
 const MODEL_PREFIX_PROVIDERS = [
+  // Codex CLI sends this bare virtual model for auto-review — keep it on OAuth Codex (#1398).
+  [/^codex-auto-review$/, "codex"],
   [/^claude-/, "anthropic"],
   [/^gemini-/, "gemini"],
   [/^gpt-/, "openai"],

tests/unit/codex-auto-review-routing.test.js

@@ -0,0 +1,44 @@
+import { describe, expect, it } from "vitest";
+
+import {
+  getDefaultModel,
+  getModelQuotaFamily,
+  getModelUpstreamId,
+  getProviderModels,
+} from "../../open-sse/config/providerModels.js";
+import { getModelInfoCore } from "../../open-sse/services/model.js";
+
+// Codex CLI's auto-review sends the bare model id "codex-auto-review". Before #1398 it fell
+// through prefix inference to the "openai" default and failed with
+// "No active credentials for provider: openai".
+describe("codex auto-review routing (#1398)", () => {
+  it("routes the bare Codex auto-review model to the OAuth Codex provider", async () => {
+    await expect(getModelInfoCore("codex-auto-review", {})).resolves.toEqual({
+      provider: "codex",
+      model: "codex-auto-review",
+    });
+  });
+
+  it("exposes Codex auto-review as a review-quota Codex model", () => {
+    const autoReview = getProviderModels("cx").find(
+      (model) => model.id === "codex-auto-review",
+    );
+
+    expect(autoReview).toBeTruthy();
+    expect(autoReview.name).toBe("Codex Auto Review");
+    expect(getModelQuotaFamily("cx", "codex-auto-review")).toBe("review");
+  });
+
+  // getModelUpstreamId strips CODEX_REVIEW_SUFFIX from unregistered "cx" ids, which would send
+  // "codex-auto" upstream. This model is not a derived review variant, so it must go out verbatim.
+  it("forwards the id upstream without stripping the -review suffix", () => {
+    expect(getModelUpstreamId("cx", "codex-auto-review")).toBe(
+      "codex-auto-review",
+    );
+  });
+
+  // Registering it must not push it to the front of the cx list — getDefaultModel takes models[0].
+  it("does not become the default Codex model", () => {
+    expect(getDefaultModel("cx")).not.toBe("codex-auto-review");
+  });
+});