You are getting the Agent execution terminated due to error in Antigravity for Mac on every run. The app accepts prompts, then instantly fails or hangs with no output.
Antigravity on Mac shows Agent execution terminated due to error on every run
The issue presents as a total block: every model selection fails, authentication looks fine, and even with MCP servers disabled you still see the same message. It often points to a stale session, corrupted local state, or a backend outage that the client does not handle gracefully.
Solution Overview
Aspect
Detail
Root Cause
Stale auth session or corrupted local cache causing generation to abort. Possible backend outage or account flag.
Primary Fix
Fully quit Antigravity, purge local state by renaming the config folder, relaunch, and re authenticate.
Try a very short prompt in Antigravity, then try the same prompt in the web workspace you normally use. If the web works but the app fails, it is likely a local state issue.
If everything fails during peak hours, confirm capacity issues and wait a short period. Guidance is here: server traffic checks.
Confirm your subscription and region access are active. If there is any account flag you can hit hard errors in the app. See account eligibility fixes.
Network quick check from Terminal to rule out blocked connectivity:
curl https://clients3.google.com/generate_204
You should see no output and an exit code of zero.
Step 2 Fully quit the app and kill stale processes
Sometimes the UI closes but helpers remain and keep corrupted state loaded.
killall Antigravity
pkill Antigravity
ps aux | grep Antigravity
If the last line still shows running Antigravity processes, repeat killall until nothing remains.
Step 3 Reset the local Antigravity cache by renaming the folder
This forces a clean schema and session reload on next launch without deleting your backup.
Open the Application Support folder:
Do the same for Logs if present so the app rebuilds fresh logs:
open "$HOME/Library/Logs"
If you later need to restore, you still have the .bak copy.
Step 4 Relaunch and re authenticate
Start Antigravity again.
Sign out in the app, then sign back in to refresh the token.
Run a simple prompt first to confirm generation succeeds.
Step 5 Keep MCP servers and tools off for the first run
Ensure MCP servers and third party tools are disabled for your first validation run.
After a clean success, re enable them one at a time and test.
Step 6 Validate with multiple models
Run a short prompt with your default model and again with a different model to confirm the core engine is healthy.
If one model fails and another works, it suggests a backend issue for that specific model. Use the working model for now and retry the other later.
Alternative Fixes and Workarounds
Do a clean reinstall without losing your backup
Move Antigravity to Trash from Applications.
Keep the Antigravity.bak folder from Step 3 for reference.
Re download the latest installer and reinstall.
Launch, sign in, and test a simple prompt.
Run Antigravity from a fresh mac user profile
Create a new mac user and install Antigravity there.
If it works in the new profile, the original profile has a conflicting setting or corrupt preference.
Rule out enterprise proxies or content filters
Temporarily disable VPN and proxy and retry.
If you must remain on a corporate network, ask IT to allow standard outbound HTTPS to Google AI endpoints and websockets.
Troubleshooting Tips
Check time and date accuracy in System Settings. Large clock drift can break token validation.
Launch the Console app and filter for Antigravity to spot client errors or repeated auth failures.
Test repeatability with a one line prompt. If short prompts work but large ones fail, it can be an attachment or context payload issue.
If the message only appears during busy periods, it is likely transient pressure on the service. Here is a quick reference for capacity related issues: high traffic guidance.
If your subscription or region is not recognized in the desktop app, review the remediation steps here: eligibility troubleshooting.
Best Practices
Keep Antigravity updated to the latest build. Many generation aborts are fixed in point releases.
Periodically clear or archive the Antigravity folder under Application Support if you see growing logs or schemas after heavy usage.
Re authenticate after long idle periods to refresh session tokens.
Enable MCP servers one at a time and confirm stability after each addition.
Use small test prompts before long runs to confirm the agent is healthy.
In most cases the Agent execution terminated due to error is resolved by a full quit, a clean local state rebuild, and a fresh sign in. If the failures persist across models after these steps, it usually points to a temporary service issue or an account flag, so keep an eye on capacity and eligibility checks and try again shortly.
We use cookies to ensure that we give you the best experience on our website. If you continue to use this site we will assume that you are happy with it.
