fix(gateway): make moderation score optional — LLM omits it in media batches

result.score was required by zod; the LLM (gemini-3.5-flash-lite via 9router)
occasionally omits it for media batches, hard-failing the whole batch parse
('Zod validation failed: expected number, received undefined' at
results[0].score). Callers already null-coalesce (result.score ?? 0) and the
parser clampScore()s it, so requiring it only caused parse failures.
Adds regression tests: media-batch without score parses (score->0), and
score-present responses still parse with the value.

services/discord-gateway/src/modules/ai-moderation/moderationSchemas.ts

@@ -23,7 +23,10 @@ export const ResultItemSchema = z.object({
   message_id: z.union([z.string(), z.number()]).transform(String),
   status: z.enum(["clean", "warn", "flagged"]),
   flags: z.array(z.string()).optional(),
-  score: z.number(),
+  // score is optional — the LLM occasionally omits it (especially media
+  // batches) and callers already null-coalesce (`result.score ?? 0`).
+  // Requiring it hard-fails the whole batch parse for one missing field.
+  score: z.number().optional(),
   analysis: z.string().nullable().optional(),
   categories: z.array(z.string()).optional(),
   severity: SeveritySchema.optional(),

services/discord-gateway/tests/moderationScoreOptional.test.ts

@@ -0,0 +1,57 @@
+import { describe, expect, it } from "vitest";
+import { parseModerationResponse } from "../src/modules/ai-moderation/moderationResponseParser.js";
+
+const NO_SCORE_BODY = JSON.stringify({
+  results: [
+    {
+      message_id: "1544590005719146557",
+      status: "clean",
+      flags: [],
+      severity: "none",
+      confidence: 1.0,
+      recommended_action: "none",
+      policy_version: "default-2026-05-30",
+      evidence: [],
+      analysis: "Normal conversation, nothing concerning.",
+    },
+  ],
+});
+
+describe("parseModerationResponse — score optional", () => {
+  it("accepts a media-batch response WITHOUT score (LLM omits it)", () => {
+    // Before the fix this threw "Zod validation failed ... expected number,
+    // received undefined" at results[0].score — the exact production error.
+    let rows: ReturnType<typeof parseModerationResponse>;
+    expect(() => {
+      rows = parseModerationResponse(NO_SCORE_BODY, ["1544590005719146557"]);
+    }).not.toThrow();
+    const row = rows![0];
+    expect(row.messageId).toBe("1544590005719146557");
+    expect(row.status).toBe("clean");
+    // clampScore(undefined, 0) coalesces to 0.
+    expect(row.score).toBe(0);
+    expect(row.flags).toEqual([]);
+    expect(row.severity).toBe("none");
+  });
+
+  it("still accepts a response WITH score", () => {
+    const rows = parseModerationResponse(
+      JSON.stringify({
+        results: [
+          {
+            message_id: "1",
+            status: "flagged",
+            flags: ["violence"],
+            score: 0.93,
+            severity: "high",
+            recommended_action: "delete",
+          },
+        ],
+      }),
+      ["1"],
+    );
+    expect(rows[0].messageId).toBe("1");
+    expect(rows[0].status).toBe("flagged");
+    expect(rows[0].score).toBeCloseTo(0.93, 5);
+  });
+});