Antigravity OAuth Rate-Limiting & Context Bloat (Claude 4.5/Gemini Pro)
Issue: VPS/Ubuntu - Antigravity OAuth Context Bloat & Extreme Latency
Environment:
* OS: Ubuntu (VPS/Headless)
* Provider: Google -> Antigravity OAuth
* Tier: AI Pro Account
* Models Affected: Claude 4.5 Opus, Gemini Pro (Gemini 3 Flash works but fails eventually too)
* Channel: Telegram Bot
* Install: Recent fresh install from scratch
The Problems:
1. Extreme Latency: Responses often take 30–40 minutes or hang indefinitely.
2. Instant Rate Limits: Hitting "Quota Exhausted" or "Rate Limit" after just 1 simple query, even with a fresh session and active AI Pro sub.
3. Context Overflow: The agent frequently reports that the context is too large. It feels like the session history isn't being summarized or pruned, causing the payload to spiral until the API times out.
4. Resilience: Once the "hang" starts, the bot becomes unresponsive until I perform a complete manual reset of the session.
What I’ve Tried:
* [x] Performed a full fresh installation.
* [x] Verified AI Pro status and API quotas.
* [x] Attempted
/compact* [x] Checked for zombie processes on the VPS.
Questions for the Team:
* Is there a specific Context Pruning/Compaction setting I should adjust for the Antigravity provider to prevent this bloat on headless Ubuntu?
* Are there known issues with OAuth session desync on VPS that cause these massive timeouts when using high-context models like Opus?
* Is there a recommended
userAgent
The AI that actually does things. Emails, calendar, home automation - all from your favorite chat app. New shell, ready to help. The lobster way.
83,815Members
Resources
Similar Threads
Was this page helpful?
