fix(usage): restore xai quota label case; update stale kiro tests

3-way unit-suite comparison (branch-pre-merge vs origin/master vs
HEAD) with git worktrees:

* ProviderLimits/utils.js: parseQuotaData 'xai' case (ab9a3c1d
  weekly/api_usage label mapping) was dropped by an EARLIER merge
  (already failing at the pre-merge branch tip) — its companion test
  xai-usage.test.js has been red since. Restored verbatim from
  ab9a3c1d; the xai usage handler + dispatch entry survived, only the
  UI label mapping was lost.
* openai-to-kiro.test.js: 19 failures were stale upstream tests —
  1fc2a81d intentionally removed the redundant top-level
  systemPrompt field (Kiro rejects it) without updating them. The
  thinking/agentic directives now ride the frozen session-start msg0
  via contentPrefix. systemPromptOf reads msg0 (or the current
  message when there is no replayed session); the cross-turn
  stability test asserts the actual cacheability contract.

Suite now: 83 failing vs 103 on origin/master baseline; zero files
fail in HEAD that did not fail pre-merge. next build passes.

src/app/(dashboard)/dashboard/usage/components/ProviderLimits/utils.js

@@ -666,6 +666,48 @@ export function parseQuotaData(provider, data) {
         }
         break;
 
+      case "xai":
+        // xAI mixes:
+        // - weekly: percentage window (used/total 0-100) from GetGrokCreditsConfig
+        // - api_usage: absolute monthly credits from /v1/billing
+        // For absolute rows, do not forward remainingCredits as `remaining`
+        // (QuotaTable treats remaining as a 0-100 percentage; same pitfall as Qoder).
+        if (data.quotas) {
+          Object.entries(data.quotas).forEach(([quotaType, quota]) => {
+            const name =
+              quotaType === "weekly"
+                ? "Weekly limit"
+                : quotaType === "api_usage"
+                  ? "Api usage"
+                  : quotaType === "monthly"
+                    ? "Api usage"
+                    : quotaType === "on_demand"
+                      ? "On-demand"
+                      : quotaType;
+
+            if (quotaType === "weekly") {
+              normalizedQuotas.push({
+                name,
+                used: quota.used || 0,
+                total: quota.total || 100,
+                remaining: quota.remaining,
+                remainingPercentage: quota.remainingPercentage ?? quota.remaining,
+                resetAt: quota.resetAt || null,
+              });
+              return;
+            }
+
+            normalizedQuotas.push({
+              name,
+              used: quota.used || 0,
+              total: quota.total || 0,
+              unit: quota.unit,
+              resetAt: quota.resetAt || null,
+            });
+          });
+        }
+        break;
+
       default:
         // Generic fallback for unknown providers
         if (data.quotas) {

tests/unit/openai-to-kiro.test.js

@@ -11,7 +11,16 @@ import { openaiToKiroRequest } from "../../open-sse/translator/request/openai-to
 
 const contentOf = (result) =>
   result.conversationState.currentMessage.userInputMessage.content;
-const systemPromptOf = (result) => result.systemPrompt || "";
+// 1fc2a81d dropped the redundant top-level systemPrompt field (Kiro rejects
+// it): thinking/agentic directives ride into the frozen session-start message
+// (msg0) via contentPrefix, or into the current message when there is no
+// session to replay. Read whichever carries them.
+const systemPromptOf = (result) => {
+  const cs = result.conversationState || {};
+  const msg0 = cs.history?.[0]?.userInputMessage?.content;
+  if (typeof msg0 === "string" && msg0) return msg0;
+  return cs.currentMessage?.userInputMessage?.content || "";
+};
 
 describe("openaiToKiroRequest", () => {
   describe("basic message conversion", () => {
@@ -568,23 +577,36 @@ describe("openaiToKiroRequest", () => {
       expect(systemPromptOf(result)).toContain("<max_thinking_length>16000</max_thinking_length>");
     });
 
-    it("keeps top-level systemPrompt stable across turns", () => {
+    it("keeps the frozen session-start directive stable across turns", () => {
+      // Upstream dropped the redundant top-level systemPrompt; cacheability
+      // now lives in the replayed msg0. Assert the directive rides msg0 and
+      // the volatile clock only enters the current turn.
+      const credentials = {
+        connectionId: "kiro-account-openai-stable",
+        rawHeaders: { "x-session-id": "hermes-session-openai-stable" },
+      };
       const first = openaiToKiroRequest(
         "claude-sonnet-4.6-thinking",
         { messages: [{ role: "user", content: "first" }] },
         true,
-        {}
+        credentials
       );
       const second = openaiToKiroRequest(
         "claude-sonnet-4.6-thinking",
         { messages: [{ role: "user", content: "second" }] },
         true,
-        {}
+        credentials
       );
 
-      expect(first.systemPrompt).toBe(second.systemPrompt);
-      expect(first.systemPrompt).not.toContain("Current time");
-      expect(first.conversationState.currentMessage.userInputMessage.content).toContain("Current time");
+      const msg0 = second.conversationState.history[0].userInputMessage.content;
+      // msg0 carries the thinking directive and is replayed byte-identical on
+      // later turns (upstream prefix includes the session-start timestamp;
+      // the cacheability contract is that it does NOT change per turn).
+      expect(msg0).toBe(first.conversationState.currentMessage.userInputMessage.content);
+      expect(msg0).toContain("<max_thinking_length>");
+      const cur = second.conversationState.currentMessage.userInputMessage.content;
+      expect(cur).toContain("second");
+      expect(cur).toContain("Current time");
     });
 
     it("replays frozen msg0 for explicit Kiro sessions while keeping current time fresh", () => {