Repository navigation
Description
Area
apps/server (Antigravity ACP adapter)
Summary
In T3 Code 0.1.9, the Antigravity ACP adapter sets promptInactivityTimeout: "15 minutes" (the only provider with this watchdog) and only pauses it while an ACP-visible foreground tool or subagent is active (AcpAdapterV2.promptInactivityPaused).
Antigravity's native schedule tool (step_type = 132) is not surfaced as an active ACP tool call, yet agy_acp_server keeps session/prompt open while waiting for the timer. As a result:
- Any native
scheduletimer>= 15 minutes(or a recurring cron) deterministically hits the 15-minute inactivity watchdog and killsagy_acp_server:Antigravity sent nothing for 15 minutes, so T3 Code stopped the turn and restarted the agent. Send a message to continue. - Killing the ACP runtime mid-timer orphans the
schedulestep inantigravity-acp/conversations/<sessionId>.dbatsteps.status = 2(CORTEX_STEP_STATUS_RUNNING) and protobuf field 4 =0x20 0x02insidestep_payload. - On every subsequent turn after
session/load,agy_acp_servernever reconciles the orphanedstatus = 2row—even if the agent callsmanage_task(Action="kill")(which returns"Task ... cancelled."without updating SQLite). Every future turn delivers its full assistant reply (step_type = 15,status = 3), never settlessession/prompt, and fails 15 minutes later with the same error in an infinite loop.
Steps to reproduce
- Start a T3 Code thread using the Antigravity provider (e.g.
gemini-4-neon). - Ask the agent to wait 20 minutes before continuing (or run a workflow where the agent calls native
schedule(DurationSeconds=1200, Prompt="...")and ends its tool calls to wait for the reactive wakeup). - Wait 15 minutes.
- Observe T3 Code kill and restart the Antigravity runtime with:
Antigravity sent nothing for 15 minutes, so T3 Code stopped the turn and restarted the agent. Send a message to continue. - Send any short follow-up message in the same thread (e.g.,
"status?"). - Observe that the agent replies immediately, but the turn stays
runningwith no active tools and fails again exactly 15 minutes later with the same error—and repeats on every subsequent turn.
Expected behavior
- Long waits (CI/CD, remote jobs, user-requested delays) should not crash Antigravity threads at 15 minutes, or native
scheduleshould be disabled in favor of T3 Code's ownwatch_pull_request/schedule_taskMCP tools. - Restarting the ACP runtime after a watchdog timeout should never leave the session DB in a permanently wedged state where every future completed reply hangs for 15 minutes and fails.
Actual behavior
- Observed 6 consecutive watchdog failures across 2 separate Antigravity threads (4 consecutive failures on one thread after a 27m
schedule(DurationSeconds=1620)call; 2 consecutive failures on another thread after a 20mschedule(DurationSeconds=1200)call). - Inspecting
<GEMINI_HOME>/antigravity-acp/conversations/<sessionId>.dbon both threads confirmed a single orphanedstep_type = 132(schedule) row frozen atstatus = 2(RUNNING), withstep_payloadprefix0884012002(field 1 =132, field 4 =2), even after multiple restarts andmanage_task(Action="kill"). - Flipping that row's
statuscolumn and protobuf field 4 byte from2(RUNNING) to6(CANCELED, prefix0884012006) while the DB is idle unsticks the session. - (Separately, a 7th 15-minute failure in the same 48-hour window was caused by two silent upstream
502"The model is currently unreachable."stepsstep_type = 17, matching Antigravity silent 502s wait out the full 15m inactivity timer and fail to parse in AntigravityTurnOutcome #33583.)
Impact
Major degradation or frequent failure (any long-running CI/CD, scrape, deploy, or delayed-resume workflow on Antigravity/Gemini threads repeatedly fails and permanently wedges the thread until a fresh session is created or the SQLite DB is manually patched).
Version
- T3 Code Desktop (Beta):
0.1.9 - Antigravity ACP server (
agy_acp_server.par):1.4.5(darwin_arm64)
Environment
- OS: macOS
26.6.2(arm64) - Provider / Model: Antigravity (
gemini-4-neon)
Suggested fixes (T3 Code side)
- Disable native
scheduleinantigravitySessionMeta(quickest high-leverage fix): T3 Code already passes_meta.antigravity.disabledTools: ["run_workflow"]on Antigravity 1.4.5+. Adding"schedule"todisabledTools(or surfacingscheduleas an active ACP tool that pausespromptInactivityPaused) prevents Antigravity from entering untracked in-prompt timer waits when T3 Code already provideswatch_pull_requestandschedule_task. - Prevent or recover from orphaned
status = 2steps after watchdog teardown: WhenretireInactivePrompttears downagy_acp_server, or beforesession/load, detect/reconcile orphanedstatus = 2(RUNNING)schedule/ background steps (or avoid failing an already-deliveredstep_type = 15, status = 3reply). - Make
promptInactivityTimeoutconfigurable per provider instance (or pause it when Antigravity reports active waiting/thinking), as also requested in [Bug]: Antigravity sent nothing for 15 minutes, so T3 Code stopped the turn and restarted the agent. Send a message to continue. #34160.
Workaround
- Configure an Antigravity
PreToolUsehook in<GEMINI_HOME>/config/hooks.jsonwith"matcher": "schedule"that blocksCronExpressionandDurationSeconds > 600(10m), steering the agent to usewatch_pull_request,schedule_task, or<= 10mchained check-in hops. - For already-wedged threads, once idle, back up
<sessionId>.dband update orphanedstep_type = 132, status = 2rows (bothsteps.statusandstep_payloadprefix0884012002->0884012006) to6(CANCELED), or start a fresh thread.
Related issues
- [Bug]: Antigravity sent nothing for 15 minutes, so T3 Code stopped the turn and restarted the agent. Send a message to continue. #34160 (same 15-minute watchdog error + sticky per-session hang across restarts)
- [Bug]: Antigravity ACP background tasks remain stuck in RUNNING state after manage_task kill or timer expiry #25599 (orphaned
steps.status = 2persisting across restarts andmanage_task(kill)) - Antigravity silent 502s wait out the full 15m inactivity timer and fail to parse in AntigravityTurnOutcome #33583 (silent upstream 502s hitting the same 15-minute watchdog)
Report prepared with Antigravity (gemini-4-neon).
Activity
juliusmarminge commentedon Nov 22, 2026
This is the same problem as #34160, so I'm closing it in favour of that one. If it is different, say how and it will be reopened.