429 Agent execution terminated due to error

Google Antigravity
1

Subject: [Urgent] Persistent HTTP 429 RESOURCE_EXHAUSTED Across ALL Models on Antigravity 2.0 v2.8.1 (Google AI Pro Plan)

Body:

Dear Antigravity Support / Engineering Team,

I am writing to report a critical and persistent HTTP 429 RESOURCE_EXHAUSTED error that has completely blocked all Agent executions for 7 consecutive days on my active Google AI Pro subscription.

Environment & Client Info

  • OS: Windows

  • Client Version: Antigravity 2.0 (v2.8.1)

  • Plan: Google AI Pro (Active, 100% quota remaining)

  • Duration: Issue has persisted for 7 days without recovery

Latest Error Log & Telemetry

  • Trajectory ID: 158ff07f-cae8-4997-8634-501ae4f1a5..

  • Sherlog TraceID: 0x9b9dc9152d90cd..

  • X-Cloudaicompanion-Trace-Id: 9b9dc9152d90cd..

  • Server Header: ESF

  • Timestamp: Wed, 19 Aug 2026 06:33:48 GMT

JSON

{
  "error": {
    "code": 429,
    "message": "Resource has been exhausted (e.g. check quota).",
    "status": "RESOURCE_EXHAUSTED"
  }
}

Verified Details & Troubleshooting Completed

  1. Quotas are 100% Intact:

    • Under Settings → Models, 100% remaining is displayed for both Weekly Limit and Five Hour Limit across Gemini, Claude, and GPT models.

  2. Universal Failure Across All Models:

    • I have tested every single available model (Gemini 3.7 Flash, Claude 3.7 Sonnet, GPT-4o, etc.). Every attempt immediately fails with the identical HTTP 429 status code before any generation starts, indicating a gateway-level session/routing block (cloudcode-pa / ESF layer) rather than a specific model’s quota limit.

  3. Local Cleanup & Reinstallation Performed:

    • Completely clean-reinstalled Antigravity to v2.8.1.

    • Wiped all local cache and credential directories (%APPDATA%\Antigravity, ~/.antigravity, and Windows Credential Manager tokens).

    • Verified network routing with full TUN proxy and anti-DNS leak configurations.

Request

Because all quotas are at 100% and every model triggers the same instant 429, my account appears stuck in a backend “ghost quota lock” or deadlocked session on the Companion Gateway.

Could you please use the Trace-ID (9b9dc9152d90cd..) and Trajectory ID (158ff07f-cae8-4997-8634-501ae4f1a5..) to investigate and manually reset/clear the cached routing session associated with my account?

Thank you for your assistance.

2

I investigated the 429 error through other AI agents and determined that it is unrelated to Google’s backend services. I tried logging into my account on someone else’s Windows machine to use Antigravity, and found that it worked normally for Q&A, but it was unable to retain previous conversation context.

First, log out of Antigravity and close the client.

Next, clear the following:
%USERPROFILE%.gemini
%APPDATA%\antigravity

Then, delete the Windows environment variable CLOUD_CODE_URL.

Finally, restart Windows and it should be resolved.

After launching Antigravity, you need to use a new working directory on Windows.

3

+1, experiencing the exact same issue on an active Google AI Pro plan with 100% quota intact across all models.

  • OS: Windows x64
  • CLI Version: 1.1.17
  • Antigravity 2.0 Version: 2.9.1
  • IDE Version: 2.5.5
  • Plan: Google AI Pro (Active, 100% quota)
  • Trajectory ID: aa6231d1-4988-4d29-90fa-1eea2f46b2..
  • Trace IDs: 0xa2690e6faf057593, 0xe6d5466f5d7bd8..
  • Error IDs: 157af031-d895-464d-b0c2-4d1758f5977d-1, 78ad3acc-e3e0-4755-b4f4-584423fddda1-1, 1a1b0392-d51c-4921-aeed-a7c954a167a5-1

Local caches, credential entries, and config directories have been completely wiped, but the companion gateway immediately terminates agent execution with 429 RESOURCE_EXHAUSTED at the pre-invocation context summarization stage before any model generation occurs.

Please flush the deadlocked session / routing cache for my account on the backend as well.

Diagnostic Logs (ls-main & Language Server) 
2026-08-21 14:07:16.907 [error] [LS Main stderr] ERROR: logging before google.Init: W0821 14:07:16.907148 117 user_management.go:170] Failed to get CAIC project: key not found
2026-08-21 14:07:17.739 [error] [LS Main stderr] ERROR: logging before google.Init: I0821 14:07:17.737876 117 http_helpers.go:228] URL: [https://cloudcode-pa.googleapis.com/v1internal:loadCodeAssist](https://cloudcode-pa.googleapis.com/v1internal:loadCodeAssist) Trace: 0x79198f6dc0a25..
2026-08-21 14:07:23.951 [error] [LS Main stderr] ERROR: logging before google.Init: I0821 14:07:23.951023 803 retry.go:40] GenerateFullResponseWithRetry: attempt 1 failed (RESOURCE_EXHAUSTED (code 429): Resource has been exhausted (e.g. check quota).), retrying in 635.990459ms
2026-08-21 14:07:24.780 [error] [LS Main stderr] ERROR: logging before google.Init: I0821 14:07:24.780204 803 retry.go:40] GenerateFullResponseWithRetry: attempt 2 failed (RESOURCE_EXHAUSTED (code 429): Resource has been exhausted (e.g. check quota).), retrying in 1.303278873s
2026-08-21 14:07:26.360 [error] [LS Main stderr] ERROR: logging before google.Init: E0821 14:07:26.359763 803 errorreport.go:223] agent executor error: pre-invocation hook: context summarization: generating user intent: RESOURCE_EXHAUSTED (code 429): Resource has been exhausted (e.g. check quota).
4

If you are experiencing the same issue as me, I have found a solution. Please refer to my reply above.