#Openclaw not working
1 messages ยท Page 1 of 1 (latest)
That Telegram message is a generic wrapper โ the real error will be in the Gateway logs.
First, please paste this output:
openclaw --version
openclaw status --all
openclaw channels status --probe
openclaw pairing list telegram
Then trigger the Telegram error once and check:
openclaw logs --follow
Donโt paste bot tokens/API keys.
One thing to look for: if logs mention PAIRING_REQUIRED, scope-upgrade, or GatewayClientRequestError: scope upgrade pending approval, fix it with:
openclaw devices list
openclaw devices approve <requestId>
openclaw gateway restart
Docs for the relevant checks: https://docs.openclaw.ai/help/faq and https://docs.openclaw.ai/channels/troubleshooting ๐ฆ
OpenClaw 2026.4.23 (a979721)
Config warnings:
- plugins.entries.slack: plugin disabled (not in allowlist) but config is present
- plugins.entries.openclaw-engram: plugin disabled (memory slot set to "memory-core") but config is present
Those two config warnings are probably not the Telegram failure โ theyโre โstale config present but plugin disabledโ warnings.
The next useful bit is the actual runtime error. Please run:
openclaw status --all
openclaw channels status --probe
openclaw pairing list telegram
Then in another terminal:
openclaw logs --follow
Send one message to your Telegram bot until it shows the error, then paste the log lines from that moment.
If you see anything like PAIRING_REQUIRED, scope-upgrade, or pending approval, run:
openclaw devices list
openclaw devices approve <requestId>
openclaw gateway restart
Telegram troubleshooting docs: https://docs.openclaw.ai/channels/troubleshooting ๐ฆ
Further output:
"Config warnings:
- plugins.entries.slack: plugin disabled (not in allowlist) but config is present
- plugins.entries.openclaw-engram: plugin disabled (memory slot set to "memory-core") but config is present
โ
Gateway reachable. - Telegram default: enabled, configured, running, connected, in:2m ago, mode:polling, bot:@Merv1Bot, token:config, groups:unmentioned, works, audit failed
Tip: status --deep adds gateway health probes to status output (requires a reachable gateway).
Config warnings:
- plugins.entries.slack: plugin disabled (not in allowlist) but config is present
- plugins.entries.openclaw-engram: plugin disabled (memory slot set to "memory-core") but config is present
08:23:23 [plugins] [lcm] Plugin loaded (enabled=true, db=/Users/macmini/.openclaw/lcm.db, threshold=0.75)
Config warnings: - plugins.entries.slack: plugin disabled (not in allowlist) but config is present
- plugins.entries.openclaw-engram: plugin disabled (memory slot set to "memory-core") but config is present
๐ฆ OpenClaw 2026.4.23 (a979721) โ If you can describe it, I can probably automate itโor at least make it funnier.
โ
โ Config warnings โโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโฎ
โ โ
โ - plugins.entries.slack: plugin disabled (not in allowlist) but config is present โ
โ - plugins.entries.openclaw-engram: plugin disabled (memory slot set to "memory-core") โ
โ but config is present "
Good news: Telegram itself is reachable here:
running, connected, mode:polling, bot:@Merv1Bot, token:config, works
So this probably isnโt a bad bot token.
Two separate things in that output:
- The plugin warnings are likely harmless cleanup noise.
groups:unmentioned, audit failedmeans your Telegram group config is asking OpenClaw to handle unmentioned group messages, but the bot/group setup may not support that. Common fixes:- In BotFather:
/setprivacyโ Disable for@Merv1Bot - remove + re-add the bot to the Telegram group
- or keep privacy mode on and mention the bot explicitly in group messages
- In BotFather:
But if youโre getting the error in a DM, the next thing to check is the agent/model error, not Telegram.
Please run this while sending one failing Telegram message:
openclaw logs --follow
Then paste the lines around the failure. Also useful:
openclaw channels status --probe --json
openclaw models status
openclaw doctor
The key bit we need is the actual exception after the Telegram message comes in โ the friendly โSomething went wrongโ message hides the real cause. Docs: https://docs.openclaw.ai/channels/telegram#troubleshooting ๐ฆ
07:49:07+00:00 info Telegram: ok (@Merv1Bot) (39ms)
07:49:07+00:00 info Agents: main (default), qwen
07:49:07+00:00 info Heartbeat interval: 2h (main)
07:49:07+00:00 info Session store (main): /Users/macmini/.openclaw/agents/main/sessions/sessions.json (284 entries)
07:49:07+00:00 info - agent:main:telegram:direct:6034907749 (5m ago)
07:49:07+00:00 info - agent:main:main:heartbeat (6m ago)
07:49:07+00:00 info - agent:main:cron:7bf409f6-8205-4391-acb6-b209f094e394 (9m ago)
07:49:07+00:00 info - agent:main:cron:7bf409f6-8205-4391-acb6-b209f094e394:run:46f699d6-edd4-49ac-8757-90f43ad1a4bb (9m ago)
07:49:07+00:00 info - agent:main:cron:7bf409f6-8205-4391-acb6-b209f094e394:run:ea50fa56-4138-4d05-bd42-35876c53f551 (29m ago)
07:49:07+00:00 warn Config warnings:
- plugins.entries.slack: plugin disabled (not in allowlist) but config is present
- plugins.entries.openclaw-engram: plugin disabled (memory slot set to "memory-core") but config is present
07:49:07+00:00 info gateway/ws {"subsystem":"gateway/ws"} โ res โ channels.status 515ms conn=207b99dfโฆ488d id=88c3f574โฆd5d8
07:49:08+00:00 info gateway/ws {"subsystem":"gateway/ws"} โ res โ doctor.memory.status 919ms conn=f3d97aedโฆ3068 id=b8f7e6e2โฆ0e0e
07:49:09+00:00 info plugins {"subsystem":"plugins"} [lcm] Plugin loaded (enabled=true, db=/Users/macmini/.openclaw/lcm.db, threshold=0.75)
07:49:10+00:00 info Run "openclaw doctor --fix" to apply changes.
That paste is from status/doctor, not the failing Telegram turn โ thereโs no inbound Telegram message or agent exception in it yet.
What we know now:
- Telegram token is good:
Telegram: ok (@Merv1Bot) - Gateway is reachable
- The failing session is likely
agent:main:telegram:direct:6034907749 - Doctor found fixable config/state issues:
Run "openclaw doctor --fix"
Iโd do this next:
openclaw doctor --fix
openclaw gateway restart
Then in Telegram send:
/new
If it still errors, run this and leave it open:
openclaw logs --follow
Then send a simple hi to the Telegram bot and paste the log lines that appear immediately after that. Also paste:
openclaw models status
At this point Telegram looks healthy, so Iโm expecting the real failure is probably model/auth/session runtime rather than the Telegram channel itself.
running "/new" in telegram gives me this message "openclaw doctor --fix
openclaw gateway restart"
sorry
Ah โ those two lines are terminal commands, not Telegram commands.
Run these on the Mac mini / machine where OpenClaw is installed:
openclaw doctor --fix
openclaw gateway restart
Then go back to Telegram and send:
/new
If it still gives the โSomething went wrongโ message, leave this running in Terminal:
openclaw logs --follow
Then send hi to the Telegram bot and paste the new log lines that appear right after that. Thatโll show the actual error. ๐ฆ
it gives me this message "โ ๏ธ Something went wrong while processing your request. Please try again, or use /new to start a fresh session."
No worries at all โ that was easy to misread.
Run those in Terminal on the OpenClaw machine:
openclaw doctor --fix
openclaw gateway restart
Then test Telegram with:
/new
hi
If it still fails, run this in Terminal and paste what appears when you send hi:
openclaw logs --follow
Yep, that means the Telegram wrapper is still hiding the real error.
Please do this exact sequence:
- Open Terminal and run:
openclaw logs --follow
- Leave that running.
- In Telegram, send:
hi
- Copy/paste the new Terminal log lines that appear right after sending
hi.
Also paste:
openclaw models status
The Telegram bot is connected, so the next suspect is model/auth/session runtime โ the logs will tell us which one.
07:52:17+00:00 error diagnostic {"subsystem":"diagnostic"} lane task error: lane=main durationMs=100 error="Error: Failed to inspect sandbox image: failed to connect to the docker API at unix:///Users/macmini/.orbstack/run/docker.sock; check if the path is correct and if the daemon is running: dial unix /Users/macmini/.orbstack/run/docker.sock: connect: no such file or directory"
07:52:17+00:00 error diagnostic {"subsystem":"diagnostic"} lane task error: lane=session:agent:main:telegram:direct:6034907749 durationMs=100 error="Error: Failed to inspect sandbox image: failed to connect to the docker API at unix:///Users/macmini/.orbstack/run/docker.sock; check if the path is correct and if the daemon is running: dial unix /Users/macmini/.orbstack/run/docker.sock: connect: no such file or directory"
07:52:17+00:00 warn model-fallback/decision {"subsystem":"model-fallback/decision"} {"event":"model_fallback_decision","tags":["error_handling","model_fallback","candidate_failed"],"runId":"095c265d-a20e-4417-a060-403d2dba6451","decision":"candidate_failed","requestedProvider":"anthropic","requestedModel":"claude-sonnet-4-6","candidateProvider":"anthropic","candidateModel":"claude-sonnet-4-6","attempt":1,"total":2,"errorPreview":"Failed to inspect sandbox image: failed to connect to the docker API at unix:///Users/macmini/.orbstack/run/docker.sock; check if the path is correct and if the daemon is running: dial unix /Users/macmini/.orbstack/run/docker.sock: connect: no such file or directory","errorHash":"sha256:34a7fbd86785","nextCandidateProvider":"openai","nextCandidateModel":"gpt-5.5","isPrimary":true,"requestedModelMatched":true,"fallbackConfigured":true} model fallback decision
07:52:17+00:00 error diagnostic {"subsystem":"diagnostic"} lane task error: lane=main durationMs=91 error="Error: Failed to inspect sandbox image: failed to connect to the docker API at unix:///Users/macmini/.orbstack/run/docker.sock; check if the path is correct and if the daemon is running: dial unix /Users/macmini/.orbstack/run/docker.sock: connect: no such file or directory"
07:52:17+00:00 error diagnostic {"subsystem":"diagnostic"} lane task error: lane=session:agent:main:telegram:direct:6034907749 durationMs=91 error="Error: Failed to inspect sandbox image: failed to connect to the docker API at unix:///Users/macmini/.orbstack/run/docker.sock; check if the path is correct and if the daemon is running: dial unix /Users/macmini/.orbstack/run/docker.sock: connect: no such file or directory"
07:52:17+00:00 error Embedded agent failed before reply: Failed to inspect sandbox image: failed to connect to the docker API at unix:///Users/macmini/.orbstack/run/docker.sock; check if the path is correct and if the daemon is running: dial unix /Users/macmini/.orbstack/run/docker.sock: connect: no such file or directory
Found it โ this is not Telegram. OpenClaw is trying to start/use the Docker sandbox, but Docker/OrbStack isnโt running or the socket path is stale:
failed to connect to the docker API at unix:///Users/macmini/.orbstack/run/docker.sock
Fastest fix: disable sandboxing for now. Run this in Terminal on the Mac mini:
openclaw config set agents.defaults.sandbox.mode off
openclaw gateway restart
Then in Telegram try:
/new
hi
If you want sandboxing enabled instead, start/install OrbStack or Docker Desktop, then verify this works in Terminal:
docker info
You can also inspect what OpenClaw thinks the sandbox config is with:
openclaw sandbox explain
Docs: https://docs.openclaw.ai/gateway/sandboxing ๐ฆ