versed saddle Apr 19, 2026, 10:58 PM

#

Hi,

I’m totally new to all of this. I bought a Hostinger VPS and installed OpenClaw from their catalog. Normally I keep trying until I fix things, but this is my 6th attempt (including reinstalling), and I keep running into a new issue every time even though I follow the same steps.

On the first day, I installed everything and it was going okay with a few issues. Then I reinstalled everything thinking I’d start fresh, but I ran into a Telegram pairing issue. Even after setting it up properly, it wouldn’t fix. It actually worked fine the first time.

Then the second time, I had a Discord setup issue, and that didn’t work either. Usually, I just ask the bot to pair things, set APIs, and approve everything.

I’ve reinstalled multiple times now, and this time I’m getting:
“You are not authorized to use this command.”

The weird part is, it was working fine before, and I did the exact same steps again.

The problem is, I rely on the bot to do everything, and I don’t really know where to manually configure things if the bot does something wrong. I also don’t know how to check or fix it myself.

I really need a proper guide for setting this up on Hostinger like:

Where to configure Discord and Telegram manually
Where to manage permissions and approvals
Where to add API keys
Where to add OpenRouter models
Where to change bot settings or names manually

Basically, I want to understand where everything is, not just rely on the bot. Also, I’m not sure how to access or use the terminal properly since it’s running through Docker.

At this point, I’m just a beginner trying to learn and use this for some basic automation. I usually don’t give up easily, but this is getting frustrating because it feels like I’m doing everything right and still failing.

Sorry for the long post, and I’d really appreciate any help.

Thank you.

wind notchBOT Apr 19, 2026, 10:58 PM

#

sick nimbusBOT Apr 19, 2026, 10:58 PM

#

To help others find answers, you can mark your question as solved via Right click solution message -> Apps -> ✅ Mark Solution

charred shadowBOT Apr 19, 2026, 11:03 PM

#

You’re not doing anything stupid here. I’m pretty sure this is a state/permissions problem, not you “doing the steps wrong”.

Also, I’d stop reinstalling for now. Reinstalls often wipe or reshuffle the exact pairing/auth state we need to inspect.

What lives where on Hostinger + Docker

If you installed OpenClaw on a Hostinger VPS, Hostinger runs it in Docker, and you manage the container from Docker Manager in hPanel.

Inside the container, OpenClaw’s persistent state lives here:

/home/node/.openclaw/

Important files there:

openclaw.json , main config

#

.env , env-backed secrets like gateway token / API keys
agents/<agentId>/agent/auth-profiles.json , stored provider auth
agents/<agentId>/sessions/ , session state

Docs:
https://docs.openclaw.ai/install/hostinger
https://docs.openclaw.ai/install/docker
https://docs.openclaw.ai/install/docker-vm-runtime#what-persists-where

Manual setup map

Discord

token/config: channels.discord.token
docs: https://docs.openclaw.ai/channels/discord

Telegram

token/config: channels.telegram.botToken

#

docs: https://docs.openclaw.ai/channels/telegram

Permissions / command auth

command allowlist: commands.allowFrom
owner-only command allowlist: commands.ownerAllowFrom
exec approvals:
- channels.discord.execApprovals.*
- channels.telegram.execApprovals.*
docs: https://docs.openclaw.ai/tools/slash-commands
https://docs.openclaw.ai/tools/exec-approvals

API keys

usually in .env or stored in the per-agent auth store
secrets docs: https://docs.openclaw.ai/gateway/secrets

OpenRouter models

model selection lives under:

#

agents.defaults.model.primary
- agents.defaults.models
key: OPENROUTER_API_KEY
docs: https://docs.openclaw.ai/concepts/models
https://docs.openclaw.ai/help/faq#how-do-i-add-models-from-other-providers-like-openrouter-or-zai

Bot name / settings

actual Discord bot name/avatar is changed in the Discord Developer Portal
actual Telegram bot name/username is changed in BotFather
OpenClaw-side reply prefix is messages.responsePrefix

Why you’re seeing “You are not authorized to use this command.”

That message usually means command authorization is blocking you, not that Discord/Telegram is broken.

Most common causes:

#

commands.allowFrom is set, and your user is not in it.
You are using an owner-only command, but commands.ownerAllowFrom does not include you.
Reinstall/reset wiped pairing or allowlist state.

So if it worked before and broke after reinstall, I’d suspect missing allowlist/owner config first.

How to access the terminal

If Hostinger gives you a VPS terminal/SSH, do this:

docker ps
docker exec -it <your-openclaw-container> sh

Then inside the container run:

#

openclaw --version
openclaw status --all
openclaw gateway status
openclaw channels status --probe
openclaw config file
openclaw pairing list telegram
openclaw devices list
openclaw approvals get

For logs:

openclaw logs --follow

#

If you don’t know the container name, docker ps will show it.

What I need from you next

Please paste these:

openclaw --version
openclaw status --all
openclaw gateway status
openclaw channels status --probe

And also tell me:

the exact command that replies with You are not authorized to use this command.
whether you’re using Hostinger OpenClaw VPS or 1-click managed OpenClaw

#

if possible, a screenshot or paste of the Docker Manager logs

Once you send that, I can usually tell you exactly which setting is blocking you.

versed saddle Apr 19, 2026, 11:12 PM

#

Can you help me go through the step-by-step process, one step at a time, until the next one?

charred shadowBOT Apr 19, 2026, 11:13 PM

#

Yep, we can do that. One step at a time, no rushing.

Step 1

Open your Hostinger VPS terminal (or SSH into the VPS) and run this:

docker ps

Then look at the NAMES column, pick the OpenClaw container name, and run:

docker exec -it <THAT_CONTAINER_NAME> sh
openclaw --version

#

Paste the output here.

If any part of that fails, paste the exact error and I’ll adjust the next step.
Also, please don’t reinstall again while we debug this.

versed saddle Apr 19, 2026, 11:13 PM

#

CONTAINER ID IMAGE COMMAND CREATED STATUS PORTS NAMES
bcb16a5eb2c8 ghcr.io/hostinger/hvps-openclaw:latest "/entrypoint.sh node…" 6 hours ago Up 3 hours 0.0.0.0:46022->46022/tcp, [::]:46022->46022/tcp openclaw-urge-openclaw-1
666f785f701e traefik:latest "/entrypoint.sh --ap…" 7 hours ago Up 7 hours traefik-vjda-traefik-1
root@srv1584299:~#

charred shadowBOT Apr 19, 2026, 11:14 PM

#

Nice, that helps. Your OpenClaw container is:

openclaw-urge-openclaw-1

Step 2

Run this next:

docker exec -it openclaw-urge-openclaw-1 sh
openclaw --version

Then paste the version output here.

#

If openclaw --version says command not found, paste that exact error and stop there.

versed saddle Apr 19, 2026, 11:21 PM

#

OpenClaw 2026.4.12 (1c0672b)

charred shadowBOT Apr 19, 2026, 11:21 PM

#

Perfect, thanks.

Step 3

Still inside the container, run:

openclaw status --all

Paste the full output here.

If it’s very long, paste the part starting at Gateway and the part showing Channels.

versed saddle Apr 19, 2026, 11:25 PM

#

Channels
┌──────────┬─────────┬────────┬──────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────┐
│ Channel │ Enabled │ State │ Detail │
├──────────┼─────────┼────────┼──────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────┤
│ Discord │ ON │ OK │ token config×2 (sha256:5e0f1b56 · len 72) · accounts 2/2 │
│ Telegram │ ON │ OK │ token config (sha256:1a0ab647 · len 46) · accounts 1/1 │
└──────────┴─────────┴────────┴─────────────────────────────────────────────────────────

charred shadowBOT Apr 19, 2026, 11:25 PM

#

Good news, this already tells us a lot: Discord and Telegram tokens are loaded correctly. So the problem is probably authorization/pairing, not a broken bot token.

Step 4

Run this next:

openclaw config get commands --json

Paste the output here.

Also, if you can, tell me the exact slash command that gives you:

You are not authorized to use this command.

That part will matter a lot.

versed saddle Apr 19, 2026, 11:27 PM

#

openclaw config get commands --json

{
"native": "auto",
"nativeSkills": "auto",
"text": true,
"bash": true,
"config": true,
"plugins": true,
"restart": true,
"useAccessGroups": false,
"ownerDisplay": "raw"

#

I get this with every comman:

You are not authorized to use this command.

charred shadowBOT Apr 19, 2026, 11:27 PM

#

Aha, I think we found the problem.

Your commands config has no ownerAllowFrom set.

That means owner-only commands will fail with:

You are not authorized to use this command.

So if you were using things like /config or /plugins, this is very likely the reason.

Step 5

In the Discord chat/thread with OpenClaw, run:

/id

#

(or /whoami if /id doesn’t appear)

Paste the result here.

That should give us your Discord sender ID, then the next step will be adding you to commands.ownerAllowFrom.

#

Got it, that’s useful. If every command is blocked, then this is not just the owner-only /config problem.

That points more toward Discord command authorization / allowlist / group policy.

Step 6

Back in the container, run this:

openclaw config get channels.discord --json

Paste the output here.

If it’s huge, paste the parts that mention any of these:

allowFrom

#

groups
groupPolicy
commands
accounts

versed saddle Apr 19, 2026, 11:28 PM

#

🧭 Identity
Channel: discord
User id: 1302991150763999302
Username: @versed saddle
AllowFrom: 1302991150763999302
🧭 Identity
Channel: discord
User id: 1302991150763999302
Username: @versed saddle
AllowFrom: 1302991150763999302

charred shadowBOT Apr 19, 2026, 11:29 PM

#

Perfect. That gives us your Discord user ID, and /id working confirms the bot can see you.

Step 7

Inside the container, run this exact command:

openclaw config set commands.ownerAllowFrom '["discord:1302991150763999302"]' --strict-json

Then paste the output here.

That should make you the authorized owner for owner-only commands like /config and /plugins.

versed saddle Apr 19, 2026, 11:29 PM

#

openclaw config set commands.ownerAllowFrom '["discord:1302991150763999302"]' --strict-json

🦞 OpenClaw 2026.4.12 (1c0672b) — Ah, the fruit tree company! 🍎

Config overwrite: /data/.openclaw/openclaw.json (sha256 f0655039e1579b5f31aaf5653202980fe861d7d3ef87ec998efc874392d7d4e7 -> d126e22233d7d628bd0652a878c6fdf619dfa54da1c8ed24ab07ff34d3dc4edc, backup=/data/.openclaw/openclaw.json.bak)
Updated commands.ownerAllowFrom. Restart the gateway to apply.

charred shadowBOT Apr 19, 2026, 11:30 PM

#

Nice, that worked. Also, this tells us your Hostinger image is storing config at:

/data/.openclaw/openclaw.json

That’s useful to know.

Step 8

Restart the OpenClaw container so the new owner setting takes effect.

If you are still inside the container, first leave it:

exit

#

Then on the VPS host run:

docker restart openclaw-urge-openclaw-1

Paste the output here, then we’ll test a command.

versed saddle Apr 19, 2026, 11:30 PM

#

/data/.openclaw/openclaw.json

sh: 7: /data/.openclaw/openclaw.json: Permission denied

charred shadowBOT Apr 19, 2026, 11:31 PM

#

Yep, that’s just because the file path is not a command. You accidentally tried to run the config file itself.

No problem.

Step 8 again

From the VPS host shell, run this exact command:

docker restart openclaw-urge-openclaw-1

Paste the output here.

Do not type /data/.openclaw/openclaw.json by itself. That file is just where the config is stored.

versed saddle Apr 19, 2026, 11:31 PM

#

I don't know where the section is

charred shadowBOT Apr 19, 2026, 11:31 PM

#

You don’t need a special section, just the normal VPS terminal prompt where you saw:

root@srv1584299:~#

That is the right place.

Step 8, simplified

At that prompt, type exactly this:

docker restart openclaw-urge-openclaw-1

Then press Enter and paste whatever it prints.

#

If you are still stuck inside the container first, type:

exit

until you get back to the root@srv... prompt, then run the restart command.

versed saddle Apr 19, 2026, 11:32 PM

#

Can't I just click restart button ?

charred shadowBOT Apr 19, 2026, 11:32 PM

#

Yes, absolutely. That’s the same thing, and honestly easier here.

Step 8, easier version

In Hostinger Docker Manager, click Restart for:

openclaw-urge-openclaw-1

Wait about 20 to 30 seconds for it to come back up, then tell me when that’s done.

After that, we’ll test whether the authorization error is gone.

versed saddle Apr 19, 2026, 11:35 PM

#

Still getting: You are not authorized to use this command on my server ID 1495157727146545194

charred shadowBOT Apr 19, 2026, 11:35 PM

#

Ah, that changes it. Thanks, that’s actually a really important clue.

If it works here but fails on your own server, then your user is authorized, but your Discord server/guild is probably not allowed in the bot’s Discord config.

So now we need to inspect the Discord server access settings.

Step 9

Back in the container, run:

openclaw config get channels.discord --json

Paste the output here.

If it’s long, paste only the parts that mention:

#

groups
groupPolicy
allowFrom
accounts
requireMention

That should tell us why your server 1495157727146545194 is being blocked.

versed saddle Apr 19, 2026, 11:37 PM

#

openclaw config get channels.discord --json
openclaw: command not found

charred shadowBOT Apr 19, 2026, 11:37 PM

#

Yep, that just means you’re back on the VPS host, not inside the OpenClaw container.

No worries, we can do it in one line so it’s easier.

Step 9 again

Run this exact command from the VPS prompt:

docker exec -it openclaw-urge-openclaw-1 sh -lc 'openclaw config get channels.discord --json'

Paste the output here.

That runs the openclaw command inside the container for you.

versed saddle Apr 19, 2026, 11:38 PM

#

docker exec -it openclaw-urge-openclaw-1 sh -lc 'openclaw config get channels.discord --json'
{
"enabled": true,
"allowBots": true,
"actions": {
"channels": true,
"messages": true,
"threads": true,
"pins": true,
"reactions": true,
"search": true,
"channelInfo": true,
"memberInfo": true,
"roleInfo": true,
"roles": true,
"events": true,
"presence": true,
"moderation": true,
"polls": true,
"stickers": true,
"emojiUploads": true,
"stickerUploads": true,
"permissions": true,
"voiceStatus": true
},
"agentComponents": {
"enabled": true
},
"voice": {
"autoJoin": [],
"enabled": true
},
"accounts": {
"custom-1": {
"ackReactionScope": "all",
"allowFrom": [
"1302991150763999302"
],
"allowBots": true,
"agentComponents": {
"enabled": true
},
"groupPolicy": "allowlist"
},
"default": {
"token": "OPENCLAW_REDACTED",
"allowFrom": [
"GJ7BRSVP",
"1302991150763999302",
"1495157727146545194",
"1495312653156089867"
],
"groupPolicy": "allowlist"
}
},
"allowFrom": [
"1302991150763999302",
"1495157727146545194",
"1495312653156089867",
"1495312602283249806"
],
"capabilities": [
"1302991150763999302"
],
"defaultAccount": "1302991150763999302",
"defaultTo": "1302991150763999302",
"contextVisibility": "all",
"dangerouslyAllowNameMatching": true,
"dmPolicy": "pairing",
"dms": {},
"token": "OPENCLAW_REDACTED",
"groupPolicy": "allowlist"
}
root@srv1584299:~#

charred shadowBOT Apr 19, 2026, 11:39 PM

#

Okay, I see the issue now.

You have two Discord accounts configured, and the important part is:

default account allows your server 1495157727146545194
custom-1 account does not
custom-1 is set to groupPolicy: "allowlist"

So if your bot in that server is the custom-1 Discord account, it will reject commands there.

Step 10

Run this exact command from the VPS prompt:

docker exec -it openclaw-urge-openclaw-1 sh -lc 'openclaw config set channels.discord.accounts.custom-1.allowFrom "[\"1302991150763999302\",\"1495157727146545194\"]" --strict-json'

#

Paste the output here.

That adds your user ID and your server ID to the custom-1 Discord account allowlist.

versed saddle Apr 19, 2026, 11:39 PM

#

docker exec -it openclaw-urge-openclaw-1 sh -lc 'openclaw config set channels.discord.accounts.custom-1.allowFrom "["1302991150763999302","1495157727146545194"]" --strict-json'

🦞 OpenClaw 2026.4.12 (1c0672b) — Like having a senior engineer on call, except I don't bill hourly or sigh audibly.

Config overwrite: /data/.openclaw/openclaw.json (sha256 061585a4603fd3281d053b7bc444f554ecc998a11d9f557fe6795f2b23fcdb82 -> ad3eb4d191af478dee4ba68ec4697eddaaebee1479505d968a475508fcb2af4e, backup=/data/.openclaw/openclaw.json.bak)
Updated channels.discord.accounts.custom-1.allowFrom. Restart the gateway to apply.
root@srv1584299:~#

charred shadowBOT Apr 19, 2026, 11:40 PM

#

Nice, that part worked too.

Step 11

Restart the OpenClaw container one more time so that new Discord allowlist applies.

You can just use the Restart button again in Hostinger Docker Manager for:

openclaw-urge-openclaw-1

When it’s back up, tell me, and then we’ll test a command on your own server.

versed saddle Apr 19, 2026, 11:41 PM

#

You are not authorized to use this command.

Still same

charred shadowBOT Apr 19, 2026, 11:42 PM

#

I found the real issue, and this one’s on me, I pointed you at the wrong Discord allowlist earlier.

For Discord servers, OpenClaw does not use allowFrom as the main server allowlist.
With groupPolicy: "allowlist", it expects your server to be listed under:

channels.discord.guilds

Right now your config has groupPolicy: "allowlist" but no guilds block, so server commands get denied.

Step 12

Run this exact command from the VPS prompt:

#

docker exec -it openclaw-urge-openclaw-1 sh -lc 'openclaw config set channels.discord.guilds "{\"1495157727146545194\":{\"requireMention\":true,\"users\":[\"1302991150763999302\"]}}" --strict-json'

Paste the output here.

That adds your server to the Discord guild allowlist and allows your user ID inside it.

versed saddle Apr 19, 2026, 11:46 PM

#

🦞 OpenClaw 2026.4.12 (1c0672b) — Claws out, commit in—let's ship something mildly responsible.

Config overwrite: /data/.openclaw/openclaw.json (sha256 1b77595876e5ec56dbdc6c221a6202b4b3eb5514ace00b1bea9e08489b252255 -> 3be0f54a938719aa2ec5b4607f8511ad990be0277cd7c873c972cd5fcdb3f835, backup=/data/.openclaw/openclaw.json.bak)
Updated channels.discord.guilds. Restart the gateway to apply.

charred shadowBOT Apr 19, 2026, 11:47 PM

#

Nice, that’s the right change.

Step 13

Restart the OpenClaw container again in Hostinger Docker Manager.
After it comes back up, go to your own server and run:

/id

Then tell me which one happens:

it shows your identity info, or
it still says You are not authorized to use this command.

We’re very close now.

versed saddle Apr 19, 2026, 11:49 PM

#

OMG it finally worked

charred shadowBOT Apr 19, 2026, 11:49 PM

#

Hell yes 🎉

Root cause was:

you were not set as an owner for owner-only commands, and
your Discord server was not in channels.discord.guilds, while Discord was running with groupPolicy: "allowlist".

So the important manual places for Discord were:

commands.ownerAllowFrom
channels.discord.guilds

Docs for this part: https://docs.openclaw.ai/channels/discord

If you want, next we can do one step at a time for either:

Telegram setup

#

API keys / OpenRouter models
a simple “where everything lives” map for Hostinger

And if this fixed the original problem, please right-click the solution message → Apps → ✅ Mark Solution so Answer Overflow can index it for the next person.

versed saddle Apr 19, 2026, 11:50 PM

#

How can i fix it in the future? Possible without coding?

charred shadowBOT Apr 19, 2026, 11:51 PM

#

Yes, absolutely. This was config, not coding.

For your Hostinger setup, think of it like this:

1. Platform settings

These are outside OpenClaw.

Discord bot name/avatar: Discord Developer Portal
Telegram bot name/username: BotFather

2. OpenClaw config

This is the main “manual fix” layer.

On your Hostinger image, the important config file is:

#

/data/.openclaw/openclaw.json

The big ones to remember:

Owner permissions
commands.ownerAllowFrom
Discord server access
channels.discord.guilds
Telegram token
channels.telegram.botToken
Discord token
channels.discord.token

#

OpenRouter model choice
agents.defaults.model.primary
agents.defaults.models

3. Container control

This is just operations, not coding.

Restart: use the Docker Manager Restart button
Logs: use Docker Manager logs, or terminal
Status: check whether channels/tokens are actually loaded

Best way to avoid this in the future

Do this first, not reinstall

Reinstalling usually hides the real issue.

#

Instead check:

docker exec -it openclaw-urge-openclaw-1 sh -lc 'openclaw status --all'
docker exec -it openclaw-urge-openclaw-1 sh -lc 'openclaw config get commands --json'
docker exec -it openclaw-urge-openclaw-1 sh -lc 'openclaw config get channels.discord --json'
docker exec -it openclaw-urge-openclaw-1 sh -lc 'openclaw config get channels.telegram --json'

If Discord says “not authorized”

Check these first:

commands.ownerAllowFrom
channels.discord.guilds
channels.discord.groupPolicy

If Telegram breaks

#

Check these first:

channels.telegram.botToken
pairing / approval state
Telegram group policy if using groups

If models/API keys break

Check these first:

.env or auth store
agents.defaults.model.primary
agents.defaults.models

Short answer

#

Yes, future fixes are possible without coding.
You mainly need:

the config file
the restart button
a few read-only openclaw config get ... commands

That’s troubleshooting, not programming.

If you want, I can make you a super simple Hostinger cheat sheet next, like:

“Discord broken → check these 3 things”
“Telegram broken → check these 3 things”
“Models broken → check these 3 things”

versed saddle Apr 19, 2026, 11:56 PM

#

How can I do it without the config code for discord?

charred shadowBOT Apr 19, 2026, 11:57 PM

#

Yes, mostly.

For Discord, you have 3 non-coding-ish options:

1. Ask the bot in plain English

Now that your permissions are fixed, this is the easiest one.

Example:

Add my Discord Server ID 1495157727146545194 to the guild allowlist

That pattern is even in the docs for Discord guild setup.
Docs: https://docs.openclaw.ai/channels/discord

2. Use the Control UI in the browser

This is the best “no config code” option.

#

Open the Control UI from Hostinger, then use:

Channels for channel status / per-channel config
Config for form-based editing
Apply + Restart

The Control UI supports view/edit config and apply with restart, so you do not have to hand-write JSON every time.
Docs: https://docs.openclaw.ai/web/control-ui

3. Use the guided wizard, not raw config

From terminal, instead of editing JSON directly, use:

openclaw configure

or section-based setup like:

#

openclaw config --section channels

That’s guided setup, not coding.
Docs: https://docs.openclaw.ai/reference/wizard

Short version

If you mean “can I avoid writing config JSON by hand?”
Yes, usually by using:

the bot in plain English
the Control UI
the guided configure wizard

If you want, I can give you the easiest future workflow for Discord only, like a 3-step “do this every time” version.

versed saddle Apr 19, 2026, 11:59 PM

#

I want ot wipe everything of this discord connection and start again, how to do it? Reset the whole setting I have on OpenClaw setting in Channels, seassions. agentsa, automation etc

charred shadowBOT Apr 19, 2026, 11:59 PM

#

Yes, you can do that, and you do not need to reinstall the whole VPS.

I’d use OpenClaw’s built-in reset, not a fresh Hostinger reinstall.

Best options

Option A, recommended

Reset OpenClaw config + credentials + sessions, then run setup again.

This is the clean “start over” option for:

channels
pairings
sessions
agents config
automation / cron config
stored auth / API creds

#

It does not wipe the whole workspace.

From the VPS terminal:

docker exec -it openclaw-urge-openclaw-1 sh
openclaw onboard --reset

Then choose:

Config + credentials + sessions

Option B, nuclear reset

If you want absolutely everything gone, including workspace:

#

docker exec -it openclaw-urge-openclaw-1 sh
openclaw onboard --reset --reset-scope full

That is the real “wipe everything” button.

My recommendation

Use Option A first.

It’s much safer, and for your case it’s probably exactly what you want.

Important warning

This will remove the stuff you set up, including things like:

#

Discord / Telegram config
stored tokens / auth
sessions
pairings
agent config
automation config

So only do it if you really want a fresh start.

Docs: https://docs.openclaw.ai/reference/wizard

If you want, I can walk you through the reset one step at a time and make sure you pick the right reset scope.

versed saddle Apr 20, 2026, 12:05 AM

#

  How do you want to hatch your bot?
│  Hatch in TUI (recommended)
 openclaw tui - ws://127.0.0.1:18789 - agent main - session main                                                                                                                                                                                                                          
 connecting | idle                                                                                                                                                                                                                                                                        
 gateway disconnected: control ui requires device identity (use HTTPS or localhost secure context) | idle                                                                                                                                                                                 
──────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────
gateway connect failed: GatewayClientRequestError: control ui requires device identity (use HTTPS or localhost secure context)

charred shadowBOT Apr 20, 2026, 12:06 AM

#

Yep, that error makes sense on a Hostinger VPS.

The fix is:

Do not choose “Hatch in TUI” here.
When it asks:

How do you want to hatch your bot?

choose:

Do this later

That’s the right choice for this kind of remote setup.

#

Why this happens

The TUI is trying to connect to the gateway over:

ws://127.0.0.1:18789

and your gateway is rejecting it because that path needs proper auth/device identity in this context.

What to do now

If you’re still stuck in the TUI screen:

Ctrl+C

#

Then run the reset/onboarding again, and when the hatch prompt appears, select:

Do this later

After that, continue the reset/setup flow normally.

If you want, I can keep going with you step by step from that exact point.

versed saddle Apr 20, 2026, 12:09 AM

#

Is it safe to rset fully and reconnect the same api key and bot? or should i create new API key and re add it?

#

Also, Should I install via terminal or Hostinger Docker Catalog?

charred shadowBOT Apr 20, 2026, 12:10 AM

#

Yes, it’s usually safe to fully reset OpenClaw and reconnect the same bot + same API key.

You do not need new keys/tokens just because you reset OpenClaw.

Reuse the same key/token if:

the key was not leaked
the bot itself is fine
you just want a clean local reset

Create a new key/token only if:

you pasted it somewhere public
you think it was compromised
the provider-side setup is messed up and you want to rotate it on purpose

Simple rule

For your case, I’d do this:

#

same Discord bot
same Telegram bot
same API keys

Then reconnect them after the reset.

Only rotate if you have a security reason, not just because OpenClaw was reset.

Small warning

If you regenerate a token on the provider side:

the old token stops working
you must update OpenClaw with the new one

So reusing the same ones is actually the easier path.

If you want, next I can tell you which things are safe to reuse and which things should be recreated in one short checklist.

#

For you, I’d recommend Hostinger Docker Catalog, not a manual terminal install.

My recommendation

Best for beginners

Hostinger Docker Catalog / Docker Manager

Use that for:

initial install
restart
updates
basic logs

Then use the terminal only for:

checking status

#

guided fixes
running openclaw config get/set when needed

Use terminal install only if:

you want full custom Docker setup
you’re comfortable debugging Docker/CLI issues yourself
you want more control than the Hostinger template gives you

Short version

For your case:

Install with Hostinger Docker Catalog
Troubleshoot with terminal when needed
Do not mix two different install methods

That’s the cleanest path.

#

Docs: https://docs.openclaw.ai/install/hostinger
Docker docs: https://docs.openclaw.ai/install/docker

If you want, I can give you my recommended beginner setup flow for Hostinger, like “do this, then this, then this” in the safest order.

versed saddle Apr 20, 2026, 12:12 AM

#

Can i pause the old ocker container and setuop a new OpenClaw frm Catalog and add it? So, I can resume old one if something goes wrong

charred shadowBOT Apr 20, 2026, 12:12 AM

#

Yes, you can, but I would not recommend that as the beginner path.

Short answer

Possible? Yes.
Best idea? Usually no.
Safer approach? Backup the old state, then reset or redeploy one instance.

Why I don’t love the “old + new container” plan

If you set up a second OpenClaw on the same VPS, you can run into:

port conflicts
two instances using the same Discord/Telegram bot
confusing state/volume mixups
“which container is the real one?” problems

#

Also, with the same bot token, you really do not want both instances active at the same time.

Better options

Best option

Stop the old container, back up its state, then start fresh.

That gives you a rollback without two live OpenClaws fighting each other.

Advanced option

Run a second OpenClaw only if all of these are true:

separate state/volume
separate port
only one instance active with a given bot token at a time

Important detail

#

If you do this, I’d use stop, not pause.

pause = freezes the process
stop = cleaner for this kind of switch

My recommendation

For you, I’d do this:

Back up the current OpenClaw state
Stop the old container
Set up the new one
Reuse the same bot/API keys if they were not leaked
If the new setup goes badly, restore the backup

That’s much safer than juggling two live installs.

#

If you want, I can give you the safest backup-first Hostinger flow, one step at a time.

versed saddle Apr 20, 2026, 12:15 AM

#

Then should I delete the current container/docker openclaw project and open a new one?

charred shadowBOT Apr 20, 2026, 12:16 AM

#

I would not delete it first.

My recommendation

Best first move

Use a reset inside the current OpenClaw container.

That is safer and simpler than deleting the whole Docker project.

Only delete/redeploy if:

the current container/project itself is broken
reset does not help
you already made a backup and are okay starting fully fresh

Safer order

Back up the current OpenClaw state

#

Try reset in the current container
If that still feels cursed, then stop/delete/redeploy
Keep the backup so you can roll back

Short version

Reset first
Delete later only if needed
Never delete before backup

For your case, that’s the least painful path.

If you want, I can give you the exact backup → reset → redeploy decision flow in 3 simple steps.

versed saddle Apr 20, 2026, 12:47 AM

#

Alright, Let's fully reset then. Give me the process

charred shadowBOT Apr 20, 2026, 12:49 AM

#

Okay. If you want a true full reset, let’s do it safely.

The process

Back up the current OpenClaw state first
Run full reset
During onboarding, choose Do this later when it asks about TUI/hatching
Reconnect the same bot tokens / same API keys if they’re still valid

Important

A full reset means you are wiping the current OpenClaw state, including workspace-level stuff too.

So let’s do the safety step first.

Step 1, back it up first

#

From your VPS prompt, run this exact command:

docker cp openclaw-urge-openclaw-1:/data/.openclaw ~/openclaw-backup-$(date +%Y%m%d-%H%M%S)

That copies your current OpenClaw state out of the container into your home folder on the VPS.

Paste the output here, or tell me if it returns to the prompt with no error. Then I’ll give you Step 2, the full reset command.

versed saddle Apr 20, 2026, 1:12 AM

#

Successfully copied 6.8MB to /root/openclaw-backup-20260420-011227
root@srv1584299:~#

charred shadowBOT Apr 20, 2026, 1:13 AM

#

Perfect. Backup is safe now.

Step 2, start the full reset

Run this exact command from the VPS prompt:

docker exec -it openclaw-urge-openclaw-1 sh -lc 'openclaw onboard --reset --reset-scope full'

Important when the wizard asks:

If you see:

How do you want to hatch your bot?

#

choose:

Do this later

Do not choose TUI.

Once the reset wizard starts and you get to the next screen/question, paste it here and I’ll walk you through the next step.

versed saddle Apr 20, 2026, 1:14 AM

#

Setup mode
│  ● QuickStart (Configure details later via openclaw configure.)
│  ○ Manual```

charred shadowBOT Apr 20, 2026, 1:14 AM

#

I need a little more than just base 😅

Please paste the full prompt/question you’re seeing now, or a screenshot.

For example, copy the lines right above and below base, so I can tell you exactly which option to pick next.

#

Choose:

QuickStart

That’s the right choice for your setup and it’s much easier as a beginner.

Step 3

keep QuickStart selected
press Enter

Then paste the next screen here and I’ll tell you the next choice.

versed saddle Apr 20, 2026, 1:16 AM

#

◇  Existing config detected ──────────────╮
│                                         │
│  workspace: ~/.openclaw/workspace       │
│  model: openrouter/openai/gpt-5.4-mini  │
│  gateway.mode: local                    │
│  gateway.port: 18789                    │
│  gateway.bind: loopback                 │
│                                         │
├─────────────────────────────────────────╯
│
◆  Config handling
│  ● Use existing values
│  ○ Update values
│  ○ Reset```

charred shadowBOT Apr 20, 2026, 1:16 AM

#

Choose:

Reset

That’s the correct one since we’re doing a full clean start.

Step 4

move to Reset
press Enter

Then paste the next screen.

versed saddle Apr 20, 2026, 1:17 AM

#


◇  Reset scope
│  Full reset (config + creds + sessions + workspace)
Failed to move to Trash (manual delete): ~/.openclaw/openclaw.json
Failed to move to Trash (manual delete): ~/.openclaw/credentials
Failed to move to Trash (manual delete): ~/.openclaw/agents/main/sessions
Failed to move to Trash (manual delete): ~/.openclaw/workspace
│
◇  QuickStart ─────────────────────────╮
│                                      │
│  Gateway port: 18789                 │
│  Gateway bind: Loopback (127.0.0.1)  │
│  Gateway auth: Token (default)       │
│  Tailscale exposure: Off             │
│  Direct to chat channels.            │
│                                      │
├──────────────────────────────────────╯
│
◆  Model/auth provider
│  ● Anthropic (Claude CLI + API key)
│  ○ Arcee AI
│  ○ BytePlus
│  ○ Chutes
│  ○ Cloudflare AI Gateway
│  ○ Copilot Proxy
│  ○ Custom Provider
│  ○ DeepSeek
│  ○ Fireworks
│  ○ Gemini CLI OAuth
│  ○ GitHub Copilot
│  ○ Google
│  ○ Hugging Face
│  ○ Kilo Gateway
│  ○ LiteLLM
│  ○ LM Studio
│  ○ Microsoft Foundry
│  ○ MiniMax
│  ○ Mistral AI
│  ○ Moonshot AI (Kimi K2.5)
│  ○ Ollama
│  ○ OpenAI
│  ○ OpenAI Codex
│  ○ OpenCode
│  ○ OpenRouter
│  ○ Qianfan
│  ○ Qwen Cloud
│  ○ SGLang
│  ○ StepFun
│  ○ Synthetic
│  ○ Together AI
│  ○ Venice AI
│  ○ Vercel AI Gateway
│  ○ vLLM
│  ○ Volcano Engine
│  ○ xAI (Grok)
│  ○ Xiaomi
│  ○ Z.AI
│  ○ Skip for now

charred shadowBOT Apr 20, 2026, 1:18 AM

#

Stop here for a second.

Those lines:

Failed to move to Trash (manual delete): ...

mean the wizard did not actually remove the old state.

So if you continue from this screen, it will not be a true clean reset.

Step 5

Press:

#

Ctrl+C

to cancel the wizard and go back to the shell prompt.

Then tell me once you’re back at the prompt, and I’ll give you the safe manual reset step so we really wipe it clean.

versed saddle Apr 20, 2026, 1:18 AM

#

Yeah I back

charred shadowBOT Apr 20, 2026, 1:19 AM

#

Great.

Step 6, archive the old OpenClaw state safely

Run this exact command from the VPS prompt:

docker exec -it openclaw-urge-openclaw-1 sh -lc 'mv /data/.openclaw /data/.openclaw-pre-reset-$(date +%Y%m%d-%H%M%S) && mkdir -p /data/.openclaw'

This does not delete the old state, it just moves it aside so we get a real clean slate.

If it works, it may return to the prompt with no message, and that’s fine.

Paste the output, or tell me if it just returns to the prompt.

versed saddle Apr 20, 2026, 1:20 AM

#

Let's just delete everything and start fresh

charred shadowBOT Apr 20, 2026, 1:20 AM

#

Okay. Since you already made a backup, we can do a real wipe.

I still think archive/move is safer, but if you want delete and start truly fresh, use this.

Step 6, destructive wipe

Run this exact command from the VPS prompt:

docker exec -it openclaw-urge-openclaw-1 sh -lc 'find /data/.openclaw -mindepth 1 -maxdepth 1 -exec rm -rf {} +'

This will delete the current OpenClaw state inside the container volume.

Because you already backed it up to:

#

/root/openclaw-backup-20260420-011227

you still have a rollback.

Paste the output, or tell me if it just returns to the prompt.

versed saddle Apr 20, 2026, 1:21 AM

#

No output

charred shadowBOT Apr 20, 2026, 1:21 AM

#

Perfect. No output is fine, that usually means it worked.

Step 7

Now restart the OpenClaw container from Hostinger Docker Manager, or from the VPS prompt:

docker restart openclaw-urge-openclaw-1

After it comes back up, tell me, and then I’ll give you the clean onboarding command for the fresh setup.

versed saddle Apr 20, 2026, 1:22 AM

#

docker restart openclaw-urge-openclaw-1
openclaw-urge-openclaw-1

charred shadowBOT Apr 20, 2026, 1:23 AM

#

Nice. That restart worked.

Step 8, start fresh onboarding

Run this exact command:

docker exec -it openclaw-urge-openclaw-1 sh -lc 'openclaw onboard'

When the wizard starts:

choose QuickStart
if it later asks about hatching/TUI, choose Do this later

Then paste the next screen here, and I’ll guide you through the setup one choice at a time.

versed saddle Apr 20, 2026, 1:23 AM

#

what if it's not completed deleted?

charred shadowBOT Apr 20, 2026, 1:23 AM

#

Good question. Let’s verify first instead of guessing.

Quick check

Run this exact command:

docker exec -it openclaw-urge-openclaw-1 sh -lc 'find /data/.openclaw -mindepth 1 -maxdepth 2 | sed -n "1,40p"'

What we want

No output = it’s basically clean
Any files/folders listed = something is still there

Paste the output here, then I’ll tell you whether we’re truly clean before onboarding.

versed saddle Apr 20, 2026, 1:24 AM

#

docker exec -it openclaw-urge-openclaw-1 sh -lc 'find /data/.openclaw -mindepth 1 -maxdepth 2 | sed -n "1,40p"'
/data/.openclaw/extensions
/data/.openclaw/workspace
/data/.openclaw/workspace/state
/data/.openclaw/workspace/TOOLS.md
/data/.openclaw/workspace/SOUL.md
/data/.openclaw/workspace/.git
/data/.openclaw/workspace/.openclaw
/data/.openclaw/workspace/AGENTS.md
/data/.openclaw/workspace/HEARTBEAT.md
/data/.openclaw/workspace/BOOT.md
/data/.openclaw/workspace/BOOTSTRAP.md
/data/.openclaw/workspace/USER.md
/data/.openclaw/workspace/IDENTITY.md
/data/.openclaw/workspace/memory
/data/.openclaw/workspace/MEMORY.md
/data/.openclaw/telegram
/data/.openclaw/telegram/command-hash-default-1a0ab64749a028d8.txt
/data/.openclaw/tasks
/data/.openclaw/tasks/runs.sqlite-wal
/data/.openclaw/tasks/runs.sqlite-shm
/data/.openclaw/tasks/runs.sqlite
/data/.openclaw/logs
/data/.openclaw/logs/config-audit.jsonl
/data/.openclaw/logs/config-health.json
/data/.openclaw/credentials
/data/.openclaw/openclaw.json.bak
/data/.openclaw/openclaw.json.bak.2
/data/.openclaw/identity
/data/.openclaw/identity/device.json
/data/.openclaw/identity/device-auth.json
/data/.openclaw/openclaw.json.bak.1
/data/.openclaw/devices
/data/.openclaw/devices/pending.json
/data/.openclaw/devices/paired.json
/data/.openclaw/agents
/data/.openclaw/agents/main
/data/.openclaw/canvas
/data/.openclaw/canvas/index.html
/data/.openclaw/openclaw.json
/data/.openclaw/skills
root@srv1584299:~#

charred shadowBOT Apr 20, 2026, 1:25 AM

#

Yep, that confirms it. It is not fully clean.

The big giveaways are:

openclaw.json
openclaw.json.bak*
credentials
devices
agents

So at this point, I would stop fighting this container.

My recommendation now

Create a brand-new OpenClaw project from the Hostinger catalog instead of trying to scrub this one more.

You already have a backup, so this is the cleanest path now.

#

What I’d do

Keep the backup
Stop or delete the current OpenClaw project/container
Create a brand-new OpenClaw from Hostinger Catalog
Reconnect the same Discord bot / Telegram bot / API keys if they were not leaked
Do not have both old and new using the same bot tokens at the same time

Why I’m changing the recommendation

Normally I prefer reset first.
But now we have proof that this Hostinger setup is still repopulating old state, so a fresh catalog deployment is cleaner than more surgery.

If you want, I can now give you the exact safest sequence for:

stopping the old project
creating the new Hostinger catalog project

#

reconnecting Discord and Telegram in the right order

versed saddle Apr 20, 2026, 1:26 AM

#

So delete the old dock project and add again?

charred shadowBOT Apr 20, 2026, 1:26 AM

#

Yes, at this point, that’s what I would do.

Recommended path now

Keep your backup
- do not delete:
```
/root/openclaw-backup-20260420-011227
```
Delete the old OpenClaw Docker project/container in Hostinger
Create a brand-new OpenClaw project from the Hostinger Catalog
Reconnect the same Discord bot / Telegram bot / API keys

Important

#

Before reconnecting the same bot tokens, make sure the old OpenClaw is stopped/deleted.

You do not want both old and new instances trying to use the same bot at the same time.

Short answer

Yes. Delete the old OpenClaw Docker project, then create a new one from the Hostinger Catalog.

If you want, I can now give you the exact order in Hostinger UI, one click at a time, so you don’t accidentally remove the wrong thing.

versed saddle Apr 20, 2026, 1:27 AM

#

One more question, I can anytime restore my claww every history from hostinger vps to some other vps right or if I host to my PC?

charred shadowBOT Apr 20, 2026, 1:27 AM

#

Yes, mostly yes.

If you keep the OpenClaw state folder backup, you can usually move your setup and history to:

another VPS
your own PC
another Docker host

The important part to back up

On your Hostinger setup, the key thing is the state folder:

/data/.openclaw/

That is the main thing you’d want to migrate.

#

It usually contains things like:

config
session history
auth profiles / stored credentials
workspace files
device/session state

So, short answer

Yes, your history/config can usually be restored elsewhere by moving that state folder over.

But there are a few caveats

Some things may still need to be redone on the new machine:

browser/device pairing

#

local paths
local model setup
skill dependencies / installed binaries
anything tied to the old machine environment

Also, don’t run two live instances with the same bot tokens at the same time.

Best migration pattern

Stop old instance
Copy the state folder backup
Install OpenClaw on new machine
Restore the state folder
Start OpenClaw
Re-pair/reapprove anything machine-specific if needed

So yes, your history is portable, just not always perfectly plug-and-play.

#

If you want, I can give you a simple backup/migration checklist for:

Hostinger VPS → new VPS
Hostinger VPS → your PC

versed saddle Apr 20, 2026, 1:30 AM

#

Alright. I'll delete the docker project and add new and brb here

charred shadowBOT Apr 20, 2026, 1:31 AM

#

Sounds good.

Just keep these two things in place:

Keep your backup folder
```
/root/openclaw-backup-20260420-011227
```
When the new project is up, use only the new instance with your bot tokens

Come back here when the new Hostinger OpenClaw project is ready, and I’ll help you set it up cleanly from step 1.

wind notchBOT Apr 20, 2026, 1:53 AM

#

versed saddle Apr 20, 2026, 2:11 AM

#

Alright, so I added the OpenClaw and added Claude key for now. But I want to add OpenRouter key with GMT 5.4 Mini and 5.4

charred shadowBOT Apr 20, 2026, 2:12 AM

#

Yep, you can do that.

I’d do it without hand-writing config:

add the OpenRouter API key
then pick your primary model and fallback model

One small thing, for the exact OpenRouter model IDs, I would use the exact names shown by the wizard/UI. I don’t want to guess a stale model id from memory.

Step 1

Run this first so I can give you the exact next command for the new container:

docker ps

#

Paste the output here.

Then I’ll give you the clean command to open the model setup wizard for OpenRouter.

versed saddle Apr 20, 2026, 2:14 AM

#

CONTAINER ID IMAGE COMMAND CREATED STATUS PORTS NAMES
19cd050098b5 traefik:latest "/entrypoint.sh --ap…" 14 minutes ago Up 14 minutes traefik-i3j1-traefik-1
30f776ff36de ghcr.io/hostinger/hvps-openclaw:latest "/entrypoint.sh node…" 16 minutes ago Up 16 minutes 0.0.0.0:41470->41470/tcp, [::]:41470->41470/tcp openclaw-b8wu-openclaw-1

charred shadowBOT Apr 20, 2026, 2:15 AM

#

Perfect. Your new OpenClaw container is:

openclaw-b8wu-openclaw-1

Step 2, open model setup

Run this exact command:

docker exec -it openclaw-b8wu-openclaw-1 sh -lc 'openclaw configure --section model'

In the wizard

choose OpenRouter
paste your OpenRouter API key

#

when it asks for models, use the exact OpenRouter model names shown by the wizard

Paste the next screen here once you get to the OpenRouter/model part, and I’ll help with the next choice.

versed saddle Apr 20, 2026, 2:17 AM

#

I added the 2 models, now how can I set as default openai/gpt-5.4-mini

charred shadowBOT Apr 20, 2026, 2:18 AM

#

Small but important detail:

If you added those models through OpenRouter, then the default usually should be the OpenRouter model ID, not plain:

openai/gpt-5.4-mini

That plain ID would be for the OpenAI provider, not OpenRouter.

Two different things

Change just this chat/session: use /model
Change the global default: set agents.defaults.model.primary

Let’s verify the exact model IDs first

#

Run this:

docker exec -it openclaw-b8wu-openclaw-1 sh -lc 'openclaw config get agents.defaults.models --json'

Paste the output here.

Then I’ll tell you the exact command to set the correct one as the global default.

versed saddle Apr 20, 2026, 2:22 AM

#

``docker exec -it openclaw-b8wu-openclaw-1 sh -lc 'openclaw config get agents.defaults.models --json'
{
"anthropic/claude-opus-4-6": {
"alias": "Claude Opus 4.6"
},
"anthropic/claude-sonnet-4-6": {
"alias": "Claude Sonnet 4.6"
},
"anthropic/claude-haiku-4-5": {
"alias": "Claude Haiku 4.5"
},
"openrouter/auto": {
"alias": "OpenRouter"
},
"openrouter/openai/gpt-5.4-mini": {},
"openrouter/openai/gpt-5.4": {},
"openrouter/moonshotai/kimi-k2.5": {}```

charred shadowBOT Apr 20, 2026, 2:22 AM

#

Perfect, now we know the exact ID.

Because you added it through OpenRouter, the default should be:

openrouter/openai/gpt-5.4-mini

If you want:

default / primary = GPT 5.4 Mini
fallback = GPT 5.4

run this exact command:

docker exec -it openclaw-b8wu-openclaw-1 sh -lc 'openclaw config set agents.defaults.model "{\"primary\":\"openrouter/openai/gpt-5.4-mini\",\"fallbacks\":[\"openrouter/openai/gpt-5.4\"]}" --strict-json'

#

That gives you:

primary: openrouter/openai/gpt-5.4-mini
fallback: openrouter/openai/gpt-5.4

If you want, after that I can give you one quick command to verify the default was saved correctly.

versed saddle Apr 20, 2026, 2:25 AM

#

Done, Now I wanna add Temegram. Added the API. got the telegram id and pair code, how to add?

charred shadowBOT Apr 20, 2026, 2:25 AM

#

Nice, you’re close.

If you already have the Telegram pair code, the next step is to approve the pairing.

Do this

Run this exact command, replacing PAIR_CODE with the actual code you got:

docker exec -it openclaw-b8wu-openclaw-1 sh -lc 'openclaw pairing approve telegram PAIR_CODE'

Optional, if you want to check pending codes first

docker exec -it openclaw-b8wu-openclaw-1 sh -lc 'openclaw pairing list telegram'

#

After approving

Go back to Telegram and send the bot a normal message like:

hi

and it should start replying.

If you want, paste the pair code command output here and I’ll confirm whether it went through correctly.

versed saddle Apr 20, 2026, 2:47 AM

#

How can I start a new session like with a new name and different session? Also how can I access the old session? I added her identity and type / new and those chats are gone

charred shadowBOT Apr 20, 2026, 2:49 AM

#

/new starts a fresh current session. It does not usually delete the old one, but it does stop using it in that chat.

Important distinction

If you mean same running OpenClaw instance

/new = new current session
old session is usually still in state/history
easiest way to see old sessions is the Control UI Sessions list

If you mean the old Hostinger project you deleted

those chats are not in the new project
they only exist in your backup now

Best way to keep separate sessions with names

#

For a beginner, the easiest method is:

Use separate Discord threads

one thread = one conversation/session
give the thread a name yourself
that’s your “named session”

That is much easier than trying to juggle hidden session keys.

Docs also support thread/session binding tools like:

/focus <target>
/unfocus

but I would not start there unless you need it. Separate threads are simpler.

#

Short answer

/new = fresh session
old session is not necessarily deleted
in the same instance, use the Sessions UI to find older sessions
in a newly redeployed instance, old chats are only in your backup
easiest “named sessions” = use separate Discord threads

If you want, I can show you the simplest workflow for keeping one “main assistant” thread and one separate “project” thread so you don’t lose track again.

versed saddle Apr 20, 2026, 2:50 AM

#

Can I get back to old chats?

charred shadowBOT Apr 20, 2026, 2:50 AM

#

Yes, but it depends which “old chats” you mean.

1. Old chats from before you deleted the old Hostinger project

Those are not in the new live bot now.

But they may still be recoverable from your backup:

/root/openclaw-backup-20260420-011227

So yes, you can get them back, but by restoring the backup, not from the current fresh bot.

2. Old chats from the current live instance, before `/new`

Those are usually still in the session store, and /new just switched away from them.

In that case, the easiest way back is through the Sessions UI / Control UI.

#

Short answer

Deleted old project chats → recover from backup
Current bot + /new chats → likely still in sessions

If you want, I can help you with one specific recovery path:

recover chats from the backup, or
find chats from the current instance via sessions.

versed saddle Apr 20, 2026, 2:51 AM

#

The chat was there then I used /new

charred shadowBOT Apr 20, 2026, 2:51 AM

#

Got it. In that case, it is probably still there.

/new usually starts a new current session, it does not erase the previous one.

The catch

There usually isn’t a simple “undo /new” chat command.

Best way to get it back

Use the Control UI and look in the Sessions list.

You should be able to find the older session from just before you ran /new.

Short answer

Yes, probably recoverable
No, /new usually did not delete it
Use Sessions UI / Control UI to find the older session

#

Best practice for next time

If you want separate conversations you can come back to easily, use:

separate Discord threads, or
export important chats before resetting

That’s much easier than using /new in the same conversation.

If you want, I can walk you step by step through finding that older session in the Control UI.

versed saddle Apr 20, 2026, 2:52 AM

#

Yes please

charred shadowBOT Apr 20, 2026, 2:53 AM

#

Yep. We’ll do it one step at a time.

Step 1

Open your OpenClaw Control UI from Hostinger.

The easiest way is usually:

Hostinger hPanel
your OpenClaw project
click Open

Once the OpenClaw page is open in your browser, tell me what you see.

If it asks for a gateway token, or says pairing required, tell me that exactly and I’ll guide you through it.

versed saddle Apr 20, 2026, 2:53 AM

#

Im in the chat interface

wind notchBOT Apr 20, 2026, 2:53 AM

#

charred shadowBOT Apr 20, 2026, 2:53 AM

#

Perfect.

Step 2

Look for Sessions in the Control UI navigation, usually in the sidebar or top navigation.

Click Sessions.

Then tell me:

whether you can see a list of sessions, or
whether you don’t see Sessions anywhere

If you don’t see it, I’ll help you find the right place.

versed saddle Apr 20, 2026, 2:54 AM

#

I see only 1 session named agent:main:main

charred shadowBOT Apr 20, 2026, 2:55 AM

#

Ah, okay, that tells us a lot.

If you only see:

agent:main:main

then /new likely reset the main session in place instead of creating a second separate session entry.

So in this case, I don’t think there’s another older session for the UI to switch back to. I’m sorry, I think that old OpenClaw session is probably gone as a live session.

What that means

/new on the main chat likely replaced/reset agent:main:main
because of that, there isn’t a second old session showing in Sessions
so there’s nothing in the UI to reopen

#

For next time, safest workflow

If you want to keep old chats accessible:

Use separate Discord threads
or use /export-session before /new
avoid using /new in the one main conversation if you want to preserve that session history

Short answer

I think that specific old chat is gone as an OpenClaw session if it was in agent:main:main and /new replaced it.

If you want, I can give you a simple session workflow so this doesn’t happen again:

one main thread
one project thread
one scratch thread

versed saddle Apr 20, 2026, 2:55 AM

#

How can i start a new session but in different session way . not replacing the main or other ones

charred shadowBOT Apr 20, 2026, 2:56 AM

#

The easiest way is:

Don’t use `/new` in your main chat

Instead:

Use a separate Discord thread for each session

So your setup becomes:

main thread = your normal ongoing assistant
project thread = one separate project/session
scratch thread = testing / temporary stuff

That way you get separate conversations without replacing the main one.

Best simple workflow

#

Main chat

Keep this for your normal/default assistant.

New session

Create a new Discord thread and talk to OpenClaw there.

Give the thread a name like:

Website project
Telegram setup
Ideas / scratch

That thread name becomes your human-readable “session name”.

Advanced option

If your Discord thread bindings are enabled, commands like these exist:

#

/focus
/unfocus

But honestly, for you, I’d keep it simple and just use separate threads.

Short answer

If you want a new session without replacing main, do this:

create a new Discord thread
talk to OpenClaw there
do not use /new in the main thread

That’s the safest beginner workflow.

If you want, I can give you a 3-thread setup suggestion you can copy exactly.

versed saddle Apr 20, 2026, 3:10 AM

#

I generated the Discord Bot Token. Where to add now?

charred shadowBOT Apr 20, 2026, 3:11 AM

#

Easiest way, don’t hand-edit JSON.

Add it through the channels setup wizard

Run this:

docker exec -it openclaw-b8wu-openclaw-1 sh -lc 'openclaw configure --section channels'

Then:

choose Discord
paste the bot token
save/apply

If you’re asking where it lives manually

#

The config key is:

channels.discord.token

But I’d use the wizard, not raw config.

If you want, after you paste the token and get the next screen, send me that screen and I’ll tell you the next choice.

versed saddle Apr 20, 2026, 3:18 AM

#

I added it through the Add new container and added the name DISCORD_BOT_TOKEN and token. It's showing configured. Good right?

charred shadowBOT Apr 20, 2026, 3:19 AM

#

Yes, that is the right env var name. 👍

DISCORD_BOT_TOKEN

So if Hostinger shows it as configured, that part is good.

One important thing

If you added it after the container was already running, do a restart/redeploy once so OpenClaw actually picks it up.

Then verify it

Run:

docker exec -it openclaw-b8wu-openclaw-1 sh -lc 'openclaw channels status --probe'

#

That will tell us whether Discord is actually loading correctly.

Also remember

A correct token alone is not the whole setup. You still need:

the bot invited to your Discord server
applications.commands scope
and, if your server is restricted, the server added to the Discord allowlist

So yes, the token part sounds right.
Next step is just verifying Discord is actually connected.

versed saddle Apr 20, 2026, 3:20 AM

#

- Discord default: enabled, configured, running, connected, bot:@Miko, token:env, intents:content=limited, works
- Telegram default: enabled, configured, running, mode:polling, bot:@Claw_Miko_by_Wrakeeb_Bot, token:config, works```

charred shadowBOT Apr 20, 2026, 3:21 AM

#

Yep, that’s good. 🎉

Both channels are up:

Discord: connected, running, works
Telegram: running, works

So the token/config side is fine now.

Next real test for Discord

In your own Discord server, try:

/id

#

If that works there, your Discord setup is live.

Small note

This part:

intents:content=limited

means Discord message-content access is limited. The bot can still work, but if you want smoother normal message handling, it’s worth checking Message Content Intent in the Discord Developer Portal later.

But for now, the big thing is:

your Discord and Telegram are both successfully connected.

If you want, next I can help you do the final Discord server allowlist / test message step so you know it’s fully usable in your server.

versed saddle Apr 20, 2026, 3:21 AM

#

- Discord default: enabled, configured, running, connected, bot:@Miko, token:env, intents:content=limited, works
- Telegram default: enabled, configured, running, mode:polling, bot:@Claw_Miko_by_Wrakeeb_Bot, token:config, works```

charred shadowBOT Apr 20, 2026, 3:22 AM

#

Yep, that output is good. Both are connected and working.

Next step

In your own Discord server, try:

/id

If it replies there, Discord is working end to end.

If it does not reply there, then the next thing to fix is the server allowlist, not the token.

versed saddle Apr 20, 2026, 3:25 AM

#

It's showing

I said: I already set my Discord bot token in config. Please finish Discord setup with User ID <1302991150763999302> and Server ID <1495157727146545194>.

Reply:

I can’t finish that from here right now: the config file is unreadable to the current process, so I can’t safely patch Discord settings.

What I found:

Discord config path is channels.discord
Current config read fails with EACCES on /data/.openclaw/openclaw.json
Fix on the machine, then I can complete it:

chown 1000 "/data/.openclaw/openclaw.json"
restart the gateway
After that, I can apply the Discord user/server setup with:

User ID: 1302991150763999302
Server ID: 1495157727146545194

charred shadowBOT Apr 20, 2026, 3:25 AM

#

Yep, that reply makes sense.

It means the bot token is fine, but the bot cannot edit the config file because /data/.openclaw/openclaw.json has the wrong permissions.

Do this

Step 1, fix ownership

Run:

docker exec -it openclaw-b8wu-openclaw-1 sh -lc 'chown 1000 /data/.openclaw/openclaw.json'

Step 2, restart OpenClaw

Run:

#

docker restart openclaw-b8wu-openclaw-1

Step 3

Then ask the bot again, but use the IDs without angle brackets:

I already set my Discord bot token in config. Please finish Discord setup with User ID 1302991150763999302 and Server ID 1495157727146545194.

If you want, we can also skip the bot and I can give you the manual terminal commands right after the restart.

versed saddle Apr 20, 2026, 3:30 AM

#

She said: Done. Discord is set up with your User ID and Server ID.

charred shadowBOT Apr 20, 2026, 3:30 AM

#

Nice, that’s what we wanted 🎉

That means:

your Discord token is loaded
your user ID is allowlisted
your server ID is allowlisted

Final test

Go to your Discord server and try:

/id

or just send a normal test message if your setup is meant to respond there.

#

If it replies there, your Discord setup is properly working end to end.

And if this whole thread helped, please right-click the solution message → Apps → ✅ Mark Solution so Answer Overflow can index it for the next person.

versed saddle Apr 20, 2026, 3:32 AM

#

Me: Approve this Discord pairing code: L9RX2YVV

She: I can’t approve that pairing code from this session right now.

charred shadowBOT Apr 20, 2026, 3:33 AM

#

Yep, that’s fine, some pairing approvals won’t go through from chat.

Do it from the terminal instead.

Approve the Discord pairing code

Run this exact command:

docker exec -it openclaw-b8wu-openclaw-1 sh -lc 'openclaw pairing approve discord L9RX2YVV'

If you want to confirm pending Discord pairings first, run:

docker exec -it openclaw-b8wu-openclaw-1 sh -lc 'openclaw pairing list discord'

#

After approving, try the Discord chat again.

If you want, paste the output here and I’ll confirm it worked.

versed saddle Apr 20, 2026, 3:36 AM

#

She's giving me a new code again when texting

charred shadowBOT Apr 20, 2026, 3:36 AM

#

Yep, that’s normal in pairing mode.

If you keep messaging before the approval is completed, it can issue a new code and the older one may no longer be the active one.

Do this now

Step 1

List the current pending Discord pairing:

docker exec -it openclaw-b8wu-openclaw-1 sh -lc 'openclaw pairing list discord'

Step 2

Approve the latest/current code it shows.

Example:

#

docker exec -it openclaw-b8wu-openclaw-1 sh -lc 'openclaw pairing approve discord NEW_CODE_HERE'

Important

After that, don’t keep sending new test messages immediately.
Approve the newest code first, then send one test message.

Paste the output of openclaw pairing list discord here if you want, and I’ll tell you exactly which code to approve.

versed saddle Apr 20, 2026, 3:54 AM

#

Alright all seems like and the connections step is done. Now the config ones.

I set the activation to ⚙️ Group activation set to always. on a testing channel on the server but still getting no response. Only @ works. How can i set it to always reply without @ and also how to set for specific channel

charred shadowBOT Apr 20, 2026, 3:55 AM

#

That’s the key part:

/activation always by itself is not enough for Discord server channels.

Discord guild messages are still controlled by:

channels.discord.guilds.<SERVER_ID>.requireMention

So if it still only replies when @mentioned, that means Discord is still mention-gated.

To allow replies without @ on the whole server

Ask the bot exactly this:

#

Allow my agent to respond on this server without having to be @mentioned.

That matches the Discord docs pattern.

For only one specific channel

You need a channel-level override under your server config, basically:

server is allowlisted
that specific channel is allowlisted
requireMention: false for that channel

So the idea is:

#

channels.discord.guilds.<SERVER_ID>.channels.<CHANNEL_ID>.allow = true
channels.discord.guilds.<SERVER_ID>.channels.<CHANNEL_ID>.requireMention = false

Why it’s happening now

Because Discord server messages are mention-gated by default, and /activation always did not fully override that guild/channel config.

Easiest next step

If you want the whole server to reply without @, ask the bot with that sentence above.

If you want only one testing channel, send me the channel ID and I’ll give you the exact command/text for that specific channel.

versed saddle Apr 20, 2026, 3:56 AM

#

But I can also set it when I want by doing the /activation always/metion right? Can't i do like that so I can just do slash and choose when I want

charred shadowBOT Apr 20, 2026, 3:57 AM

#

Yes, that’s the idea, but with one important catch:

Short answer

Yes, you can use /activation mention and /activation always to switch behavior.

But it works best only after the Discord guild/channel config is already set up correctly.

What that means

Good setup

your server is allowlisted
the channel is allowlisted
then /activation mention or /activation always can be your day-to-day toggle

Bad setup

If Discord is still hard-configured as mention-only at the guild/channel level, then /activation always may not fully override it, which is exactly what you just saw.

#

Best practical setup for you

For your testing channel:

make sure that channel is allowed in config
then use:
- /activation always when you want free replies
- /activation mention when you want it quieter again

So:

config = gives the channel permission to behave that way
/activation = your convenient toggle after that

If you want, send me the testing channel ID and I’ll give you the exact setup so that /activation can be your on/off switch there.

versed saddle Apr 20, 2026, 3:59 AM

#

⚙️ Group activation set to always.

#

I just set always but not getting any messages without the mention

charred shadowBOT Apr 20, 2026, 4:00 AM

#

Good, that means the session toggle part worked.

So if it’s still not replying without @mention after that, the remaining blocker is almost certainly the Discord guild/channel config, not /activation.

Next step

Send me the testing channel ID.

Then I’ll give you the exact setup so that channel can use:

/activation always for free replies
/activation mention when you want it quiet again

If you don’t know the channel ID, enable Discord Developer Mode, then right-click the channel and choose Copy Channel ID.

#

Yep, that confirms it.

/activation always is set, but the channel itself is still mention-gated.

So the problem is not the slash command now, it’s the Discord channel config.

What I need from you

Send me the testing channel ID.

Then I’ll give you the exact setup so that channel can reply without @mention.

If you don’t know it:

turn on Discord Developer Mode
right-click the testing channel
click Copy Channel ID

versed saddle Apr 20, 2026, 4:01 AM

#

This is the channel ID: 1495621441461420062

I want the whole channel and it's server to react like that so I can use /slash anytime to change it always to mention

charred shadowBOT Apr 20, 2026, 4:02 AM

#

Perfect, and since you want the whole server to work that way, the server ID matters more than that one channel ID.

What to do

Tell the bot exactly this:

Allow my agent to respond on this server without having to be @mentioned.

That is the right server-wide baseline.

Then how `/activation` fits in

Once the server is configured that way:

/activation always = free replies in that session/channel

#

/activation mention = quieter, only when mentioned

So yes, that’s the setup you want:

server config allows no-mention replies
/activation becomes your easy toggle whenever you want

Important point

If the server is still hard-configured as mention-only, /activation always won’t fully behave the way you want.

So the order is:

set the server to allow replies without @mention
then use /activation as your switch

Since you want the whole server, you do not need to configure only that single channel first.

#

If you want, after you send that sentence to the bot, paste the reply here and I’ll tell you if it applied the correct setting.

versed saddle Apr 20, 2026, 4:09 AM

#

Me: Allow my agent to respond on this server without having to be @mentioned.

She: Done — your Discord server can now respond without @mentioning the bot.

Now, even if I change the activation via slash to mention it doesn't change. Now replying all the tme

charred shadowBOT Apr 20, 2026, 4:10 AM

#

Right, and that’s because the server config was changed, not just the session.

So now the fix is:

If you want to stop the whole server from replying all the time

Tell the bot:

Require @mention on this server again.

That should flip the server-wide Discord config back to mention-only.

Then, if you want only your testing channel to be always-on

After that, tell the bot:

#

Allow my agent to respond in channel 1495621441461420062 without @mentioning the bot.

Important

/activation is not the thing that will undo the server-wide config change on Discord.

So the pattern is:

server config controls baseline behavior
/activation is not a reliable Discord-wide override here

If you want, do the first sentence now, paste the bot’s reply, and I’ll guide the second step so only your testing channel stays always-on.

versed saddle Apr 20, 2026, 4:12 AM

#

Alright, this done for now.

So now, I can see that anyone can add her to their Apps and server. So, should I turn off public bot option from the discord developer portal? Will it stops anything to the corrent setup?

charred shadowBOT Apr 20, 2026, 4:13 AM

#

Yes, if this is meant to be your private/personal bot, I’d turn Public Bot off.

What happens if you disable Public Bot

It should:

stop other people from inviting it to new servers
not break your current setup
not remove it from servers where it is already installed

So for your current server/setup, it should be fine.

Short recommendation

Personal/private bot → turn Public Bot = off
Bot you want others to add freely → leave it on

#

If later you want to make it public again, you can just turn it back on.

versed saddle Apr 20, 2026, 4:14 AM

#

I don't want anyone to touch it. Only I can add and removbe, so should I turn off public rbot?

charred shadowBOT Apr 20, 2026, 4:15 AM

#

Yes, in your case, turn Public Bot off.

That matches exactly what you want:

no random people adding it
no public installs
only you control where it goes

It should not break the current setup you already have.

So for a private personal bot, Public Bot = off is the right move.

versed saddle Apr 20, 2026, 4:19 AM

#

Alright, so now everything is perfectly set. Now, I wanna back it up so if I lose anytime or run into some issue then i can go back. What's the process?

charred shadowBOT Apr 20, 2026, 4:20 AM

#

Nice, that’s the right time to do it.

For your Hostinger setup, the thing you want to back up is:

/data/.openclaw

That is the important state folder.

Best backup process, safest

1. Stop OpenClaw briefly

docker stop openclaw-b8wu-openclaw-1

#

2. Copy the whole state folder out of the container

docker cp openclaw-b8wu-openclaw-1:/data/.openclaw ~/openclaw-backup-$(date +%Y%m%d-%H%M%S)

3. Start OpenClaw again

docker start openclaw-b8wu-openclaw-1

Strong recommendation

After that, also copy that backup off the VPS:

your PC
cloud storage
another VPS

#

Because if the VPS dies, a backup stored only on that same VPS dies with it.

What this backup includes

Usually the important stuff:

config
session history
auth / stored credentials
workspace
device/pairing state

Restore later

If you ever need to restore:

deploy or stop the target OpenClaw container
put your backup back into /data/.openclaw
start the container again

#

Short version

For Hostinger:

backup /data/.openclaw
keep one copy off the VPS
that is your rollback point

Docs: https://docs.openclaw.ai/install/docker-vm-runtime#what-persists-where

If you want, I can give you the next exact command to make a backup right now with your current container name.

versed saddle Apr 20, 2026, 4:21 AM

#

docker cp openclaw-b8wu-openclaw-1:/data/.openclaw ~/openclaw-backup-$(date +%Y%m%d-%H%M%S)
Successfully copied 648kB to /root/openclaw-backup-20260420-042116

Done, now how to move it to my PC>

charred shadowBOT Apr 20, 2026, 4:22 AM

#

Nice. Next, make it into one file, then download that file.

Step 1, compress it on the VPS

Run:

tar -czf /root/openclaw-backup-20260420-042116.tar.gz -C /root openclaw-backup-20260420-042116

Step 2, move it to your PC

Easiest if you use Mac/Linux terminal

Run this on your PC, not on the VPS:

#

scp root@YOUR_VPS_IP:/root/openclaw-backup-20260420-042116.tar.gz ~/Downloads/

Replace YOUR_VPS_IP with your Hostinger VPS IP.

If you’re on Windows

The easiest option is usually WinSCP:

open WinSCP
connect to your VPS with:
- host = your VPS IP
- user = root
- password = your VPS password
go to:

#

   /root/

download:
```
openclaw-backup-20260420-042116.tar.gz
```

Short version

compress on VPS
use scp or WinSCP
download the .tar.gz to your PC

If you want, send me your OS (Windows / Mac / Linux) and I’ll give you the exact easiest method for your computer.

versed saddle Apr 20, 2026, 4:23 AM

#

I use Windows

charred shadowBOT Apr 20, 2026, 4:24 AM

#

Perfect, on Windows the easiest way is WinSCP.

Step 1

If you haven’t already, first make the .tar.gz on the VPS:

tar -czf /root/openclaw-backup-20260420-042116.tar.gz -C /root openclaw-backup-20260420-042116

Step 2

Download WinSCP on your PC:
https://winscp.net/

Step 3

Open WinSCP and connect with:

File protocol: SFTP

#

Host name: your Hostinger VPS IP
User name: root
Password: your VPS password
Port: 22

Step 4

After connecting, go to:

/root/

You should see:

openclaw-backup-20260420-042116.tar.gz

#

Step 5

Drag that file from WinSCP to your PC, for example:

Downloads

That’s it. Now you have an off-VPS backup.

If you want, next I can show you the restore process too, so you know how to put this backup back onto a VPS later.

versed saddle Apr 20, 2026, 4:32 AM

#

Alright, got the file of 89 kb. It contains everything right? So emprting it next time will also have memories and everything?

charred shadowBOT Apr 20, 2026, 4:34 AM

#

Mostly yes.

Because you backed up /data/.openclaw, that file should contain the important OpenClaw-side state, including things like:

config
sessions/chat history
workspace files
memory/state files
pairings/devices
stored OpenClaw data

So yes, restoring that later should bring back your OpenClaw-side setup, memory, and history.

But not literally everything

It does not automatically include things outside that state folder, like:

#

Discord Developer Portal settings
Telegram BotFather settings
VPS-wide system config
anything stored somewhere else outside /data/.openclaw

About the small file size

If your setup is still pretty fresh, 89 KB is not weird.
A small setup with limited history can still be a valid backup.

Short answer

Yes, it should include your OpenClaw memory/history/config
No, it does not include every external platform setting on earth
restoring it later should get you most of the important stuff back

If you want, I can give you a very simple restore command set next, so you know exactly how to use that backup later.

versed saddle Apr 20, 2026, 4:35 AM

#

Yeah suee go on

charred shadowBOT Apr 20, 2026, 4:36 AM

#

Nice. Here’s the simple restore flow for later.

Restore backup to a VPS / new Hostinger OpenClaw

1. Upload your backup file back to the VPS

Using WinSCP, upload this file to:

/root/

Example:

/root/openclaw-backup-20260420-042116.tar.gz

#

2. Find the OpenClaw container name

On the VPS:

docker ps

You’ll get something like:

openclaw-xxxxx-openclaw-1

#

3. Stop the OpenClaw container

Replace the name with your real one:

docker stop YOUR_OPENCLAW_CONTAINER

4. Extract the backup on the VPS

Example:

tar -xzf /root/openclaw-backup-20260420-042116.tar.gz -C /root/

That should create a folder like:

#

/root/openclaw-backup-20260420-042116

5. Copy the backup back into OpenClaw’s state path

Replace the container name:

docker cp /root/openclaw-backup-20260420-042116/. YOUR_OPENCLAW_CONTAINER:/data/.openclaw/

6. Start the container again

#

docker start YOUR_OPENCLAW_CONTAINER

Short version

Restore later =

upload .tar.gz
docker stop
tar -xzf
docker cp ...:/data/.openclaw/
docker start

Important note

#

If the new VPS already has a setup you care about, back that up first before restoring over it.

If you want, next I can give you the same restore steps rewritten as a tiny checklist you can save in Notepad for future use.

versed saddle Apr 20, 2026, 4:45 AM

#

One more thing.

I wanna add her to another server anf that server got some ppl. They can chat with her by mentioning @ or always what rules I set i up.

Is it possible to make her only accessible by me and only I can mention her and chat and no one can unless I give them permission?

charred shadowBOT Apr 20, 2026, 4:46 AM

#

Yes, absolutely. That’s possible, and it’s a good idea for a shared server.

What you want

bot is in the server
only you can trigger it
other people can see it, but they can’t use it
later, you can add specific people one by one

The Discord setup pattern for that

Use a guild allowlist like this:

{
  channels: {
    discord: {

#

      groupPolicy: "allowlist",
      guilds: {
        "SERVER_ID": {
          requireMention: true,
          users: ["YOUR_USER_ID"],
        },
      },
    },
  },
}

What that does

groupPolicy: "allowlist" = the server is controlled by allowlists
users: ["YOUR_USER_ID"] = only your Discord user ID can trigger the bot there

#

requireMention: true = even you still need to @mention it, which is safer in a shared server

If later you want to allow someone else too

Just add their Discord user ID to the same users list.

Example:

users: ["YOUR_USER_ID", "THEIR_USER_ID"]

Short answer

Yes, you can make it so only you can use the bot in that server.
And yes, you can later give permission to specific people one by one.

If you want, send me that new server ID and I’ll give you the exact sentence/command to set it up so only you can use it.

versed saddle Apr 20, 2026, 4:49 AM

#

How can I set it up? Need step by step process.

charred shadowBOT Apr 20, 2026, 4:50 AM

#

Yep, we can do it step by step.

Step 1

Get the new Discord server ID where you want to add the bot.

If you don’t have it yet:

turn on Discord Developer Mode
right-click the server icon
click Copy Server ID

Then paste that server ID here.

After that, I’ll give you Step 2.

wind notchBOT Apr 20, 2026, 4:53 AM

#

versed saddle Apr 20, 2026, 4:54 AM

#

Server ID: 1396404029680128001

charred shadowBOT Apr 20, 2026, 4:55 AM

#

Perfect.

Step 2

In your bot chat, send this exact message:

Set up Discord server 1396404029680128001 so only my user ID 1302991150763999302 can use the bot there, and require @mention.

That should configure the new server so:

the bot is present there
only you can trigger it
people must still @mention it

Paste the bot’s reply here, and I’ll give you Step 3.

#Starter help

What lives where on Hostinger + Docker

Manual setup map

Why you’re seeing “You are not authorized to use this command.”

How to access the terminal

What I need from you next

openclaw config get commands --json

openclaw config set commands.ownerAllowFrom '["discord:1302991150763999302"]' --strict-json

/data/.openclaw/openclaw.json

1. Platform settings

2. OpenClaw config

3. Container control

Best way to avoid this in the future

Do this first, not reinstall

If Discord says “not authorized”

If Telegram breaks

If models/API keys break

Short answer

1. Ask the bot in plain English

2. Use the Control UI in the browser

3. Use the guided wizard, not raw config

Short version

Best options

Option A, recommended

Option B, nuclear reset

My recommendation

Important warning

Why this happens

What to do now

Reuse the same key/token if:

Create a new key/token only if:

Simple rule

Small warning

My recommendation

Best for beginners

Use terminal install only if:

Short version

Short answer

Why I don’t love the “old + new container” plan

Better options

Best option

Advanced option

Important detail

My recommendation

My recommendation

Best first move

Only delete/redeploy if:

Safer order

Short version

The process

Important

Step 1, back it up first

Step 2, start the full reset

Important when the wizard asks:

Step 5

Step 6, archive the old OpenClaw state safely

Step 6, destructive wipe

Step 7

Step 8, start fresh onboarding

When the wizard starts:

Quick check

What we want

My recommendation now

What I’d do

Why I’m changing the recommendation

Recommended path now

Important

Short answer

The important part to back up

So, short answer

But there are a few caveats

Best migration pattern

Step 1

Step 2, open model setup

In the wizard

Two different things

Let’s verify the exact model IDs first

If you want:

Do this

Optional, if you want to check pending codes first

2. Old chats from the current live instance, before `/new`

Don’t use `/new` in your main chat

Then how `/activation` fits in