Hi,
I believe I found the cause, or at least a very strong contributing factor, behind a macOS Antigravity/Antigravity IDE sandbox issue where terminal commands remain RUNNING / show “Empty log” even after the underlying command has completed.
I have also submitted this through Antigravity’s built-in Feedback mechanism.
Symptom
With Sandbox enabled, terminal executions cause Antigravity to launch processes like:
/usr/bin/log show --style json --predicate
(senderImagePath CONTAINS “Sandbox” OR subsystem == “com.apple.sandbox.reporting”)
AND message contains "EXEBOX ID "
In my environment these processes sometimes remain alive for many minutes at approximately 100% CPU.
They accumulate as additional agent terminal commands are executed.
At one point I had 14 concurrent log show processes, each using roughly 97–100% CPU and running for up to 30+ minutes.
At the same time, Antigravity Background Task Output frequently showed “Empty log”, or commands remained RUNNING despite the underlying process having already completed. If left for half an hour, Antigravity would finally receive the data from those tasks and continue.
A/B test
I tested the same command:
node -e ‘console.log(“AG_TEST_STDOUT”); console.error(“AG_TEST_STDERR”); process.exit(42)’
Sandbox enabled:
- EXEBOX
log showcollector appears - completion/log reporting may hang
Full Access:
- stdout returned correctly
- stderr returned correctly
- exit code 42 returned correctly
- no runaway EXEBOX
log showcollector
I reproduced the sandbox collector behavior in both Antigravity IDE and Antigravity 2.0.
Interesting detail
The log show command generated by Antigravity does not appear to contain a time bound such as:
--last 5m
or:
--start …
That is potentially expensive because macOS log show queries the Unified Log datastore.
The EXEBOX ID being searched for belongs to the execution that has just started, so old log history should not be relevant.
Workaround
macOS supports default options for log show through ~/.logrc.
I created:
~/.logrc
with:
show:
–last 10m
–no-pager
After restarting Antigravity:
- EXEBOX log collectors stopped accumulating,
- terminal commands no longer hung,
- git commands returned normally,
- npm test returned its real final exit code and output,
- CPU usage returned to normal,
- Sandbox remained enabled.
I have now tested the workaround with both Antigravity IDE and Antigravity 2.0.
Possible fix in Antigravity
If this diagnosis is correct, Antigravity could probably bound its internal sandbox log query itself, e.g.:
/usr/bin/log show --last 5m --style json --predicate ‘… EXEBOX ID …’
The exact interval could obviously be different, but the important part seems to be avoiding an effectively unbounded Unified Log query for a freshly generated EXEBOX ID.
This workaround does not disable or weaken the sandbox; it only limits how much Unified Log history log show searches.
Has anyone else who experiences the RUNNING / Empty log problem on macOS checked whether log show ... EXEBOX ID processes are accumulating?
You can check with:
ps -axo pid,ppid,stat,etime,%cpu,command |
grep ‘log show.*EXEBOX ID’ | grep -v grep
I’d be interested to know whether the same fingerprint appears on other Macs.