#Config warnings:\n- plugins.entries.google-antigravity-auth: plugin removed: google-antigravity-auth
1 messages · Page 1 of 1 (latest)
That warning and the Telegram “AI service is temporarily overloaded” usually aren’t directly related.
1) Why you see plugins.entries.google-antigravity-auth: plugin removed
It means your config still contains a plugin entry for google-antigravity-auth, but your current OpenClaw build doesn’t ship that plugin anymore, so it prints a “removed” warning during config load.
Even if you never use Google auth, you can still get the warning if the stale entry is present.
Fix: remove that config entry (or let Doctor clean it up), then restart the gateway.
2) The Telegram “AI service is temporarily overloaded”
That message is a generic surface-level error when the upstream model call fails (common causes: provider overload, rate limits, transient 5xx, etc.). If you’re using Anthropic Opus, it’s typically coming from the Anthropic side (or your failover chain).
This can appear “connected” if your model call fails and OpenClaw tries fallbacks (and you have some Google provider configured), but we need to see what it’s actually doing.
What I need from you (2 quick outputs)
Please paste:
openclaw --versionopenclaw models status
If you can add one more, it’ll pinpoint the overload cause immediately:
openclaw status --all(safe to share; it redacts tokens)
Once I see models status, I can tell you whether:
- it’s pure Anthropic overload/rate-limit, or
- you have failover enabled and it’s attempting a Google provider (and tripping over that stale plugin entry).
Gateway connection details:
Gateway target: ws://127.0.0.1:18789
Source: local loopback
Config: /home/clawdbot/.openclaw/openclaw.json
Bind: loopback
âś“ Config: /home/clawdbot/.openclaw/openclaw.json
âś“ Restart sentinel: none
! Port 18789
Port 18789 is already in use.
- pid 2517154 clawdbot: openclaw-gateway (127.0.0.1:18789)
- Gateway already running locally. Stop it (openclaw gateway stop) or use a different port.
! Tailscale: off · unknown
error: Error: spawn tailscale ENOENT
✓ Skills: 19 eligible · 0 missing · /home/clawdbot/clawd
âś“ Channel issues (none)
Pasteable debug report. Auth tokens redacted.
Troubleshooting: https://docs.openclaw.ai/troubleshooting
That plugins.entries.google-antigravity-auth: plugin removed warning is basically: your config still references a plugin that no longer exists in your installed OpenClaw version. It can show up even if you “never use” Google auth — it’s just a stale config entry.
Fix the warning (safe)
Open /home/clawdbot/.openclaw/openclaw.json and remove the config at:
plugins.entries.google-antigravity-auth
(i.e. delete that one entry / block). Then restart the gateway.
If you want the fastest confirm, just search the config for google-antigravity-auth and delete the surrounding entry.
Telegram “AI service is temporarily overloaded”
That one is almost certainly unrelated to the plugin warning. It’s a generic surface message when the upstream model call fails (Anthropic overloaded / rate-limited / transient error).
To see the real reason (429 vs 5xx vs auth), we need either:
openclaw models status
- A short log snippet right after it happens:
openclaw logs --follow(trigger the error in Telegram, then paste the relevant few lines)
Side notes from your pasted status
- Port 18789 already in use: that just means your gateway is already running (pid shown). If you were trying to start another instance, use
openclaw gateway restartinstead of start. - Tailscale ENOENT: Tailscale isn’t installed on that box; ignore unless you planned to use tailnet binding/serve.
If you paste openclaw models status, I can tell you whether you’re hitting Anthropic rate limits/overload and whether any failover is being attempted.
it looks like claude is down atm
Yeah, that’s consistent.
-
The Telegram “AI service is temporarily overloaded” is exactly what you’ll see when Anthropic is having an incident / heavy load / 5xx / rate limiting. So if Claude feels “down”, that’s the likely cause.
-
The
google-antigravity-auth plugin removedwarning is separate — it’s just a stale config entry and will keep appearing until you deleteplugins.entries.google-antigravity-authfrom youropenclaw.json(then restart). It won’t cause Claude to be down.
If you want to make your setup resilient during Claude hiccups, add a fallback model/provider so OpenClaw can fail over instead of erroring. But for right now, if Anthropic is actually degraded, the only “fix” is waiting (or switching models temporarily).
If you paste openclaw models status, I can tell you what fallback options are currently configured (if any) and whether failover is working.
Okey but i want to remove this google antigravity thing, it bugs me, what exactly do i do
Remove it from your config — the warning is just telling you “your config references a plugin that isn’t shipped anymore”.
You’ve got Config: /home/clawdbot/.openclaw/openclaw.json, so do one of these:
Option A (quick CLI)
openclaw config unset 'plugins.entries.google-antigravity-auth'
openclaw gateway restart
(plugins changes generally require a restart; restart is the cleanest way to be sure it’s gone.)
Option B (edit the file directly)
- Open
/home/clawdbot/.openclaw/openclaw.json - Search for
google-antigravity-authand delete the whole entry, e.g. something like:
{
plugins: {
entries: {
"google-antigravity-auth": { /* ... */ },
},
},
}
- Save, then:
openclaw gateway restart
Option C (Control UI)
Open http://127.0.0.1:18789 → Config tab → (Raw JSON editor if needed) → remove plugins.entries.google-antigravity-auth → save → restart if it doesn’t auto-restart.
If after doing this you still see the warning, it usually means the entry is coming from an $include’d config file — search all your config fragments for google-antigravity-auth and remove it there.
Thanks G, lets hope anthropic fixes it soon