Grok Bot local-exec still flapping on 0.57.1 (residual from #170488)

Where does the bug appear (feature/product)?

Grok Bot

Describe the Bug

On Grok Bot 0.57.1 for macOS, local-exec continues to flap: the daemon stays up, but ListMachines can return empty and its log shows repeated user-computer-provider ConnectError / ECONNRESET / DeadlineExceededError stream/poll failures.

Steps to Reproduce

Use Grok Bot with local-exec on macOS. During a turn, observe ListMachines and ~/.grokbot/local-exec-daemon.log. The daemon can remain alive for days while agent-visible machines intermittently disappear and the log accumulates the stream/poll failures.

Expected Behavior

The local-exec daemon should keep a stable provider stream/poll connection, and ListMachines should consistently show the registered Mac.

Screenshots / Screen Recordings

local-exec-daemon-redacted.txt (4.47 KB)

Operating System

MacOS

Version Information

Grok Bot 0.57.1

For AI issues: which model did you use?

Grok Bot local-exec

For AI issues: add Request ID with privacy disabled

Not available anymore (from the interrupted turn).

Additional Information

@Dean Rie — follow-up on Grok Bot Computers / local-exec still flapping after update as a separate thread per your ask.

Checked today (Tue 2026-09-22, America/New_York):

  • Residual is still happening. After the Sep 6 ~1:51pm ET mid-work self-heal on 0.43.0, I also saw at least one agent-visible ListMachines empty around Sep 10 ~11:39am ET. App is now 0.57.1. Daemon process can stay up for days, but ~/.grokbot/local-exec-daemon.log still fills with user-computer-provider ConnectError / ECONNRESET / DeadlineExceededError stream/poll failures (no “desktop ownership lost” storm in the current long-lived process — flapping form shifted).
  • Log caveat: local-exec-daemon.log has no timestamps, so I cannot isolate “2 minutes before the ~1:51pm ET drop through recovery.” Attached is a redacted recent contiguous excerpt of the same pattern under 0.57.1.
  • Request ID from that interrupted turn: I don’t have it anymore. Happy to add if I can recover it.
  • Update Track: not visible in local settings files (no Early Access / releaseChannel keys). UI is Stable vs Nightly on my build — I can confirm which I’m on if needed.
  • Hardware (PII-light): MacBook Pro M2 Pro 32GB, macOS 26.x — same as the prior thread.

Attached: local-exec-daemon-redacted.txt

Thanks.

Does this stop you from using Cursor

No - Cursor works, but with this issue

Hey @ras434, thanks for the follow-up thread and the log excerpt.

What you pasted matches a pattern we already have on our side. The helper on your Mac that keeps the connection to our servers occasionally loses its long-lived stream (a server rollout or a short network blip can both cut it). While it retries with a growing wait between attempts, the Mac can drop out of the list of computers for a minute or two before it comes back on its own. That is what the self-heal you saw looks like from our side, and it is tracked as something to fix on our end, not a problem with your setup.

One reassuring note: looking at your Mac from our side, the helper has been up continuously since the 0.57.1 update on Sep 19 and we see almost none of these drops since then. So the excerpt is most likely from before that.

To catch the next one with a time we can line up, could you try this once?

  1. Open Terminal on the Mac.
  2. Paste this and press Return, then leave the window open:
    tail -F ~/.grokbot/local-exec-daemon.log | while IFS= read -r line; do printf '%s %s\n' "$(date '+%Y-%m-%d %H:%M:%S %Z')" "$line"; done >> ~/Desktop/local-exec-daemon-timestamped.log
    
  3. Keep using Grok Bot as usual. The next time an agent says the Mac is unreachable or the computer list comes back empty, note the time (ET is fine).
  4. In that chat, use the ... menu next to a message and click Copy Request ID.
  5. Reply here with the time, the Request ID, the lines from ~/Desktop/local-exec-daemon-timestamped.log from about 2 minutes before the drop through the recovery

Thanks again.