fix(antigravity): retry transient upstream failures

Retry short-lived 5xx/capacity errors (500/502/503/504 + message
patterns) with bounded backoff capped at 15s; honor Retry-After/reset
hints and skip when wait is too long. Keep 400 non-retryable. Enable the
retry hook for 500 alongside existing 429/503.

Deduplicate sanitized Antigravity tool names before emitting the single
functionDeclarations group to avoid upstream "Tool names must be unique"
rejections.

Add Headroom size diagnostics and phantom-savings warning when reported
token delta does not shrink the outbound payload.

Co-authored-by: Cursor <cursoragent@cursor.com>

tests/unit/antigravity-retry-hook.test.js

@@ -32,8 +32,38 @@ describe("antigravity computeRetryDelay hook (D3)", () => {
     expect(await ag.computeRetryDelay(res(429), 3)).toBe(Math.min(1000 * 2 ** 3, MAX));
   });
 
-  it("503 without retry info → veto (no auto backoff)", async () => {
-    expect(await ag.computeRetryDelay(res(503), 1)).toBe(false);
+  it("503 without retry info → transient backoff", async () => {
+    expect(await ag.computeRetryDelay(res(503), 1)).toBe(2000);
+  });
+
+  it("retries Antigravity agent terminated body even when status is not 429", async () => {
+    const r = res(500, {}, { error: { message: "Agent execution terminated due to error" } });
+    expect(await ag.computeRetryDelay(r, 1)).toBe(2000);
+  });
+
+  it("retries high traffic body", async () => {
+    const r = res(500, {}, { error: { message: "Our servers are experiencing high traffic" } });
+    expect(await ag.computeRetryDelay(r, 2)).toBe(4000);
+  });
+
+  it("does not retry non-transient 400 errors", async () => {
+    const r = res(400, {}, { error: { message: "Invalid request" } });
+    expect(await ag.computeRetryDelay(r, 1)).toBe(false);
+  });
+
+  it("deduplicates sanitized tool names", () => {
+    const out = ag.transformRequest("claude-opus-4-6-thinking", {
+      request: {
+        contents: [{ role: "user", parts: [{ text: "hi" }] }],
+        tools: [{ functionDeclarations: [
+          { name: "read/file", parameters: { type: "object", properties: {} } },
+          { name: "read file", parameters: { type: "object", properties: {} } },
+          { name: "read/file", parameters: { type: "object", properties: {} } },
+        ] }],
+      },
+    }, true, { projectId: "project-1", connectionId: "conn-1" });
+
+    expect(out.request.tools[0].functionDeclarations.map(fn => fn.name)).toEqual(["read_file"]);
   });
 
   it("buildHeaders includes cached session id after transformRequest", () => {

tests/unit/base-executor-retry.test.js

@@ -85,6 +85,16 @@ describe("BaseExecutor.execute — network error retry/fallback", () => {
 });
 
 describe("BaseExecutor.execute — computeRetryDelay hook veto", () => {
+  it("only invokes computeRetryDelay when status has retry config", async () => {
+    const ex = makeExec({ baseUrl: "https://x/api", retry: { 503: { attempts: 1, delayMs: 0 } } });
+    ex.computeRetryDelay = vi.fn().mockResolvedValue(0);
+    fetchMock.mockResolvedValueOnce(res(500));
+    const out = await ex.execute({ model: "m", body: {}, stream: false, credentials: creds });
+    expect(out.response.status).toBe(500);
+    expect(ex.computeRetryDelay).not.toHaveBeenCalled();
+    expect(fetchMock).toHaveBeenCalledTimes(1);
+  });
+
   it("hook returning false skips retry (uses fallback path)", async () => {
     const ex = makeExec({ baseUrl: "https://x/api", retry: { 429: { attempts: 5, delayMs: 0 } } });
     ex.computeRetryDelay = vi.fn().mockResolvedValue(false);