Api key troubles | Friends of the Crustacean 🦞🤝 | Page 1

olive bay Apr 23, 2026, 9:35 PM

#

I have the dashboard working all for one thing, claw wont accept my api key. Its the same key i put into the setup and in the json file. idk how to fix it and i cant code. im getting this error.
Agent failed before reply: No API key found for provider "openai". Auth store: C:\Users\estro.openclaw\agents\main\agent\auth-profiles.json (agentDir: C:\Users\estro.openclaw\agents\main\agent). Configure auth for this agent (openclaw agents add <id>) or copy auth-profiles.json from the main agentDir. | No API key found for provider "openai". Auth store: C:\Users\estro.openclaw\agents\main\agent\auth-profiles.json (agentDir: C:\Users\estro.openclaw\agents\main\agent). Configure auth for this agent (openclaw agents add <id>) or copy auth-profiles.json from the main agentDir. Logs: openclaw logs --follow

zealous crystalBOT Apr 23, 2026, 9:38 PM

#

This usually means OpenClaw is reading a different auth store or environment than the one you edited, not that your key is necessarily wrong.

You do not need to code for this. First, please run these and paste the output:

openclaw --version
openclaw gateway status
openclaw models status

Also, don’t paste the actual API key.

Most likely fixes on Windows:

Put the key in:

#

%USERPROFILE%\.openclaw\.env

with exactly:

OPENAI_API_KEY=your_key_here

Restart OpenClaw:

openclaw gateway restart

Check that OpenClaw is using the main agent auth store from the error:

#

~/.openclaw/agents/main/agent/auth-profiles.json

If you are using ChatGPT/Codex login instead of a Platform API key, use this instead of API-key auth:

openclaw models auth login --provider openai-codex

Relevant docs:

Paste those 3 command outputs and I’ll tell you the exact next step.

olive bay Apr 23, 2026, 9:43 PM

#

Service: Scheduled Task (registered)
File logs: ~\AppData\Local\Temp\openclaw\openclaw-2026-04-23.log
Command: C:\Program Files\nodejs\node.exe C:\Users\estro\AppData\Roaming\npm\node_modules\openclaw\dist\index.js gateway --port 18789
Service file: ~.openclaw\gateway.cmd
Service env: OPENCLAW_GATEWAY_PORT=18789

Config (cli): ~.openclaw\openclaw.json
Config (service): ~.openclaw\openclaw.json

Gateway: bind=loopback (127.0.0.1), port=18789 (service args)
Probe target: ws://127.0.0.1:18789
Dashboard: http://127.0.0.1:18789/
Probe note: Loopback-only gateway; only local clients can connect.

Runtime: running (last run 267009, last run time 4/23/2026 4:18:06 PM)
Connectivity probe: ok
Capability: read-only

Listening: 127.0.0.1:18789
Troubles: run openclaw status

Config : ~.openclaw\openclaw.json
Agent dir : ~.openclaw\agents\main\agent
Default : openai/gpt-5.4
Fallbacks (0) : -
Image model : -
Image fallbacks (0): -
Aliases (0) : -
Configured models (0): all

Auth overview
Auth store : ~.openclaw\agents\main\agent\auth-profiles.json
Shell env : off
Providers w/ OAuth/tokens (0): -

openai effective=profiles:~.openclaw\agents\main\agent\auth-profiles.json | profiles=1 (oauth=0, token=0, api_key=1) | openai:default=(ApiKey)

OAuth/token status

none

zealous crystalBOT Apr 23, 2026, 9:45 PM

#

Nice, this helps a lot.

What your output says is:

OpenClaw is looking at the right agent dir
it does see an OpenAI profile
but that profile likely has no usable secret inside it right now

So this is probably not “wrong key”, it’s more like “saved in a way the gateway can’t actually use”.

Easiest fix on Windows

Create this file:

C:\Users\estro\.openclaw\.env

#

Put exactly this in it:

OPENAI_API_KEY=your_actual_key_here

Restart the gateway:

openclaw gateway restart

Then run:

openclaw models status --probe --probe-provider openai

#

That probe should tell us if it’s fixed, or if the stored profile is broken (missing_credential / unresolved_ref).

Also paste this too

openclaw --version

Version mismatch causes a lot of weird auth issues.

If the .env fix still doesn’t do it, next clean step is to re-add OpenAI auth instead of hand-editing JSON.

Docs:

#

Paste the probe output and version, and I’ll narrow it down from there.

olive bay Apr 23, 2026, 9:48 PM

#

how do i put that in the .env folder

olive bay Apr 23, 2026, 10:47 PM

#

[agent/embedded] embedded run agent end: runId=60079504-ae99-4f92-b199-9f2b35c595fa isError=true model=gpt-5.4 provider=openai error=401 Incorrect API key provided: sk-proj-******************************************************************************************************************************************************cPoA. You can find your API key at https://platform.openai.com/account/api-keys. rawError=401 Incorrect API key provided: sk-proj-******************************************************************************************************************************************************cPoA. You …
17:45:30 [agent/embedded] auth profile failure state updated: runId=60079504-ae99-4f92-b199-9f2b35c595fa profile=sha256:a0ebe3f799a6 provider=openai reason=auth window=cooldown reused=false
17:45:30 [agent/embedded] embedded run failover decision: runId=60079504-ae99-4f92-b199-9f2b35c595fa stage=assistant decision=surface_error reason=auth from=openai/gpt-5.4 profile=sha256:a0ebe3f799a6

zealous crystalBOT Apr 23, 2026, 10:48 PM

#

Good, this narrows it down a lot.

OpenClaw is reading a key now. The new error is from OpenAI itself:

before: “no API key found”
now: 401 Incorrect API key provided

So the problem is now the stored OpenAI credential is invalid/stale, not that OpenClaw can’t see one.

Best fix

If you want normal OpenAI API usage, do this:

Make a fresh OpenAI Platform API key here: https://platform.openai.com/api-keys
Re-run onboarding so it replaces the stored OpenAI auth profile:
```
openclaw onboard --auth-choice openai-api-key
```

#

Restart the gateway:
```
openclaw gateway restart
```

Verify:

openclaw models status --probe --probe-provider openai

Important note

Because OpenClaw already has an openai auth profile in auth-profiles.json, just adding OPENAI_API_KEY to .env may not override the bad stored key. Re-running onboarding is the cleaner fix here.

If you meant ChatGPT/Codex login instead

If you’re trying to use a ChatGPT subscription instead of API billing, use this path instead:

#

openclaw models auth login --provider openai-codex

That is a different auth route from OpenAI API keys.

If you run the probe command above and paste the output, plus:

openclaw --version

I can tell you the exact next step. Docs: https://docs.openclaw.ai/providers/openai • https://docs.openclaw.ai/gateway/authentication • https://docs.openclaw.ai/help/faq

olive bay Apr 23, 2026, 10:54 PM

#

o Model configured --------------------+
| |
| Default model set to openai/gpt-5.4 |
| |
+---------------------------------------+
Error: Cannot find module '@larksuiteoapi/node-sdk'
Require stack:

C:\Users\estro\AppData\Roaming\npm\node_modules\openclaw\dist\extensions\feishu\client-yWGpoGkZ.js

zealous crystalBOT Apr 23, 2026, 10:55 PM

#

Ah, okay, this is a different problem now.

Your API key is not the immediate blocker in this message. OpenClaw is crashing because a bundled plugin dependency is missing:

Cannot find module '@larksuiteoapi/node-sdk'

That’s the Feishu/Lark plugin runtime dep. Docs say the fix is:

openclaw --version
openclaw doctor --fix
openclaw gateway restart
openclaw status

#

Then try sending one test message again.

If openclaw doctor --fix still fails, paste its full output here. That command is specifically supposed to repair missing bundled plugin runtime deps.

Relevant docs:

Small note: native Windows works, but WSL2 is still the more stable path if Windows keeps throwing weird install/runtime issues.

If this fixes it, please right-click the solution message → Apps → ✅ Mark Solution so Answer Overflow can index it.

olive bay Apr 23, 2026, 10:58 PM

#

[bonjour] watchdog detected non-announced service; attempting re-advertise (gateway fqdn=Knee-Gear (OpenClaw)._openclaw-gw._tcp.local. host=openclaw.local. port=18789 state=probing)
17:58:12 [ws] closed before connect conn=adcafa2c-199f-47e2-af63-309d61c4c4fd peer=127.0.0.1:52430->127.0.0.1:18789 remote=127.0.0.1 fwd=n/a origin=n/a host=127.0.0.1:18789 ua=n/a code=1006 reason=n/a
17:58:12 [ws] closed before connect conn=8c51a3f4-a53b-45aa-8c83-c54736dbe7ba peer=127.0.0.1:52429->127.0.0.1:18789 remote=127.0.0.1 fwd=n/a origin=n/a host=127.0.0.1:18789 ua=n/a code=1006 reason=n/a
17:58:12 [ws] closed before connect conn=3d000125-30fc-4c7c-a140-f27983f72715 peer=127.0.0.1:52428->127.0.0.1:18789 remote=127.0.0.1 fwd=n/a origin=n/a host=127.0.0.1:18789 ua=n/a code=1006 reason=n/a
17:58:12 [ws] closed before connect conn=d9a1703d-bf5d-460e-90a4-3b8703a89948 peer=127.0.0.1:52427->127.0.0.1:18789 remote=127.0.0.1 fwd=n/a origin=n/a host=127.0.0.1:18789 ua=n/a code=1006 reason=n/a
17:58:12 [ws] closed before connect conn=0a6caa24-c548-495f-9ec8-9c4cb77c0be2 peer=127.0.0.1:52426->127.0.0.1:18789 remote=127.0.0.1 fwd=n/a origin=n/a host=127.0.0.1:18789 ua=n/a code=1006 reason=n/a
17:58:12 [ws] closed before connect conn=6f8299df-8374-44aa-95ce-51f2fec77f3b peer=127.0.0.1:52424->127.0.0.1:18789 remote=127.0.0.1 fwd=n/a origin=n/a host=127.0.0.1:18789 ua=n/a code=1008 reason=connect failed
17:58:12 [plugins] embedded acpx runtime backend ready
17:58:12 [ws] closed before connect conn=a80d7b89-5b46-434d-9d6c-fba3651e68cb peer=127.0.0.1:52432->127.0.0.1:18789 remote=127.0.0.1 fwd=n/a origin=n/a host=127.0.0.1:18789 ua=n/a code=1008 reason=connect failed
17:58:13 [ws] closed before connect conn=479bafcb-acf9-4aad-bb3f-2f9fded2842e peer=127.0.0.1:52434->127.0.0.1:18789 remote=127.0.0.1 fwd=n/a origin=n/a host=127.0.0.1:18789 ua=n/a code=1008 reason=connect failed

zealous crystalBOT Apr 23, 2026, 10:59 PM

#

Those ws closed before connect lines are mostly symptoms, not the root cause.

The real blocker is still the earlier one:

Cannot find module '@larksuiteoapi/node-sdk'

That means the gateway is trying to start, pieces reconnect, then it falls over because a bundled plugin dependency is missing. So I’d do this in order:

openclaw --version
openclaw doctor --fix
openclaw gateway restart
openclaw status --all
openclaw models status --probe --probe-provider openai

#

A few notes:

Please paste all 5 outputs
Do not paste your API key
The doctor --fix step is the one that should repair missing bundled plugin runtime deps
The 1006/1008 connect failed websocket logs can happen while the gateway is half-started/restarting

If doctor --fix completes, but models status --probe --probe-provider openai still shows a 401 Incorrect API key, then the next step is to generate a fresh OpenAI Platform API key and re-run:

openclaw onboard --auth-choice openai-api-key

Docs:

#

Paste the outputs and I’ll tell you exactly which of the two issues is still left.

olive bay Apr 23, 2026, 11:07 PM

#

🦞 OpenClaw 2026.4.21 (f788c88) — Your AI assistant, now without the $3,499 headset.

|
o
Config : ~.openclaw\openclaw.json
Agent dir : ~.openclaw\agents\main\agent
Default : openai/gpt-5.4
Fallbacks (0) : -
Image model : -
Image fallbacks (0): -
Aliases (0) : -
Configured models (0): all

Auth overview
Auth store : ~.openclaw\agents\main\agent\auth-profiles.json
Shell env : off
Providers w/ OAuth/tokens (0): -

openai effective=profiles:~.openclaw\agents\main\agent\auth-profiles.json | profiles=1 (oauth=0, token=0, api_key=1) | openai:default=sk-proj-...uyMyoMcA

OAuth/token status

none

zealous crystalBOT Apr 23, 2026, 11:08 PM

#

Good news, your API key is working now.

This line is the important one:

| openai/gpt-5.4 | openai:default (api_key) | ok |

That means OpenClaw successfully made a real OpenAI auth/model probe with your stored key. So the original “won’t accept my API key” problem is basically solved.

What’s left is likely a separate runtime issue, probably the earlier missing bundled-plugin dependency on Windows.

Next step

Run these exactly:

#

openclaw doctor --fix
openclaw gateway restart
openclaw status --all

Then send one simple test prompt in chat.

Important

If you still see this:

Cannot find module '@larksuiteoapi/node-sdk'

then the remaining problem is not your OpenAI key, it’s the missing Feishu/Lark bundled plugin runtime dep.

#

You’re also on 2026.4.21, so if that module error comes back after doctor --fix, I’d update OpenClaw to the current build using the update docs, then rerun doctor:

Paste the output of openclaw doctor --fix if it complains again, and I’ll help you pin down the last issue.

olive bay Apr 23, 2026, 11:10 PM

#

🦞 OpenClaw 2026.4.21 (f788c88) — I'm like tmux: confusing at first, then suddenly you can't live without me.

▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄
██░▄▄▄░██░▄▄░██░▄▄▄██░▀██░██░▄▄▀██░████░▄▄▀██░███░██
██░███░██░▀▀░██░▄▄▄██░█░█░██░█████░████░▀▀░██░█░█░██
██░▀▀▀░██░█████░▀▀▀██░██▄░██░▀▀▄██░▀▀░█░██░██▄▀▄▀▄██
▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀
🦞 OPENCLAW 🦞

T OpenClaw doctor
|
o State integrity -------------------------------------------------------------------------+
| |
| - OAuth dir not present (~.openclaw\credentials). Skipping create because no |
| WhatsApp/pairing channel config is active. |
| - Found 1 orphan transcript file in ~.openclaw\agents\main\sessions. |
| These .jsonl files are no longer referenced by sessions.json, so they are not part of |
| any active session history. |
| Doctor can archive them safely by renaming each file to *.deleted.<timestamp>. |
| Examples: probe-openai-92e55362-298d-4d28-b848-185e6d9964d6.jsonl |
| |
+-------------------------------------------------------------------------------------------+
|
o Doctor changes ----------------------------------------------------------------------+
| |
| - Archived 1 orphan transcript file in ~.openclaw\agents\main\sessions as .deleted |
| timestamped backups. |
| |
+---------------------------------------------------------------------------------------+
|
o Security ---------------------------------+
| |
| - No channel security warnings detected. |
| - Run: openclaw security audit --deep |
| |
+--------------------------------------------+
|
o Skills status ------------+
| |
| Eligible: 6 |
| Missing requirements: 46 |
| Blocked by allowlist: 0 |
| |
+----------------------------+
|
o Plugins ------+
| |
| Loaded: 58 |
| Imported: 58 |
| Disabled: 41 |
| Errors: 0 |
| |
+----------------+
|
o
Agents: main (default)
Heartbeat interval: 30m (main)
Session store (main): C:\Users\estro.openclaw\agents\main\sessions\sessions.json (1 entries)

agent:main:main (24m ago)
gateway connect failed: GatewayClientRequestError: scope upgrade pending approval (requestId: 95cdfd11-071b-4063-a69d-683f6af90d3c)
|
o Device pairing ---------------------------------------------------------------------------+
| |
| - Pending scope upgrade 95cdfd11-071b-4063-a69d-683f6af90d3c for |
| gateway:device.pair.list |
| (492e802ba3d2fe6905ddeef2ad6f07b5d2cc0e81584acdd81a07005f2b6817ba): approved scopes |
| [operator.read], requested scopes [operator.pairing]. Review with openclaw devices |
| list, then approve with openclaw devices approve 95cdfd11-071b-4063-a69d-683f6af90d3c. |
| |
+--------------------------------------------------------------------------------------------+
|
— Doctor complete.

zealous crystalBOT Apr 23, 2026, 11:11 PM

#

This is actually good news, your OpenAI key is working now.

The proof is this line:

| openai/gpt-5.4 | openai:default (api_key) | ok |

So the API key problem is basically gone. You do not need to keep re-entering the key.

What Doctor is showing now is a different issue: a pending device scope upgrade.

Run these:

openclaw devices list

#

openclaw devices approve 95cdfd11-071b-4063-a69d-683f6af90d3c
openclaw gateway restart

Then try the assistant again.

A couple notes:

gateway connect failed: scope upgrade pending approval is not an OpenAI key error
the orphan transcript cleanup is harmless
if the old API-key error still shows after restart, send openclaw status --all

Since you're on 2026.4.21, if weird module/plugin errors come back after this, I'd update next:

#

If this fixed it, please right-click the solution message → Apps → ✅ Mark Solution so Answer Overflow can index it.

olive bay Apr 23, 2026, 11:13 PM

#

🦞 OpenClaw 2026.4.21 (f788c88) — WhatsApp automation without the "please accept our new privacy policy".

|
o
[openclaw] Failed to start CLI: Error: gateway closed (1006 abnormal closure (no close frame)): no close reason
Gateway target: ws://127.0.0.1:18789
Source: local loopback
Config: C:\Users\estro.openclaw\openclaw.json
Bind: loopback

Possible causes:

Gateway not yet ready to accept connections (retry after a moment)
TLS mismatch (connecting with ws:// to a wss:// gateway, or vice versa)
Gateway crashed or was terminated unexpectedly
Run openclaw doctor for diagnostics.
at Object.onClose (file:///C:/Users/estro/AppData/Roaming/npm/node_modules/openclaw/dist/call-Cl0YkVGh.js:251:10)
at WebSocket.<anonymous> (file:///C:/Users/estro/AppData/Roaming/npm/node_modules/openclaw/dist/client-D-67f4NF.js:314:23)
at WebSocket.emit (node:events:509:28)
at WebSocket.emitClose (C:\Users\estro\AppData\Roaming\npm\node_modules\openclaw\node_modules\ws\lib\websocket.js:263:12)
at emitErrorAndClose (C:\Users\estro\AppData\Roaming\npm\node_modules\openclaw\node_modules\ws\lib\websocket.js:1047:13)
at ClientRequest.<anonymous> (C:\Users\estro\AppData\Roaming\npm\node_modules\openclaw\node_modules\ws\lib\websocket.js:886:5)
at ClientRequest.emit (node:events:509:28)
at emitErrorEvent (node:_http_client:109:11)
at Socket.socketErrorListener (node:_http_client:593:5)
at Socket.emit (node:events:509:28)

zealous crystalBOT Apr 23, 2026, 11:14 PM

#

Yep, this is no longer an API key issue.

Your earlier probe showed OpenAI auth is ok. This new error means the gateway itself is closing/crashing before the CLI can talk to it:

gateway closed (1006 abnormal closure)
Gateway target: ws://127.0.0.1:18789

Try this exact sequence:

openclaw gateway status
openclaw gateway restart
openclaw gateway status
openclaw logs --follow

#

Then, after restart, try one simple command like:

openclaw status

What I’m looking for

If gateway status says stopped or unhealthy, the gateway is the blocker
If logs --follow shows the old Feishu/Lark error again:
```
Cannot find module '@larksuiteoapi/node-sdk'
```
then that missing bundled-plugin dep is still the crash cause
The 1006 line itself is just “the gateway disappeared”, not the root cause

Important note

You’re on 2026.4.21, which is a bit behind current. If restart still crashes, I’d update next, then rerun doctor:

#

Paste:

openclaw gateway status
the first real error from openclaw logs --follow

and I’ll help you pin the crash down.

olive bay Apr 23, 2026, 11:16 PM

#

🦞 OpenClaw 2026.4.21 (f788c88) — Automation with claws: minimal fuss, maximal pinch.

zealous crystalBOT Apr 23, 2026, 11:17 PM

#

Yep, this is the blocker now.

Your OpenAI auth is fine. The CLI is being blocked because this device only has operator.read, and it’s asking for a scope upgrade.

If this is your own machine/device, approve the pending request:

openclaw devices approve 26637641-69e2-4f82-ab73-3ee7c9fb4c3d
openclaw gateway restart
openclaw status

Why: the output shows a pending scope upgrade from operator.read to the broader operator scopes OpenClaw needs.

If it still fails after approval, run:

#

openclaw devices list
openclaw gateway status

and paste that.

Docs:

If this fixed it, please right-click the solution message → Apps → ✅ Mark Solution.

olive bay Apr 23, 2026, 11:20 PM

#

[bonjour] watchdog detected non-announced service; attempting re-advertise (gateway fqdn=Knee-Gear (OpenClaw)._openclaw-gw._tcp.local. host=openclaw.local. port=18789 state=probing)
18:19:02 [ws] closed before connect conn=4c8790e1-f33f-4610-ace0-2c871060dc52 peer=127.0.0.1:52514->127.0.0.1:18789 remote=127.0.0.1 fwd=n/a origin=n/a host=127.0.0.1:18789 ua=n/a code=1006 reason=n/a
18:19:02 [ws] closed before connect conn=546bc835-d804-4f3e-b911-298aa542b0df peer=127.0.0.1:52513->127.0.0.1:18789 remote=127.0.0.1 fwd=n/a origin=n/a host=127.0.0.1:18789 ua=n/a code=1006 reason=n/a
18:19:02 [ws] closed before connect conn=f0445555-8951-46b9-808f-f379629fa19f peer=127.0.0.1:52512->127.0.0.1:18789 remote=127.0.0.1 fwd=n/a origin=n/a host=127.0.0.1:18789 ua=n/a code=1006 reason=n/a
18:19:02 [ws] closed before connect conn=dcdad4a8-f2ad-4448-9ff8-8a0b8cc503e0 peer=127.0.0.1:52511->127.0.0.1:18789 remote=127.0.0.1 fwd=n/a origin=n/a host=127.0.0.1:18789 ua=n/a code=1006 reason=n/a
18:19:02 [ws] closed before connect conn=5c527786-0d66-45c9-b6cb-e4d484a48827 peer=127.0.0.1:52510->127.0.0.1:18789 remote=127.0.0.1 fwd=n/a origin=n/a host=127.0.0.1:18789 ua=n/a code=1006 reason=n/a
18:19:02 [ws] closed before connect conn=60042984-6ef4-4711-8213-3576fe26e490 peer=127.0.0.1:52508->127.0.0.1:18789 remote=127.0.0.1 fwd=n/a origin=n/a host=127.0.0.1:18789 ua=n/a code=1008 reason=connect failed
18:19:02 [plugins] embedded acpx runtime backend ready
18:19:03 [ws] closed before connect conn=d025304f-6243-4d4a-9315-41fb2d5bcae4 peer=127.0.0.1:52516->127.0.0.1:18789 remote=127.0.0.1 fwd=n/a origin=n/a host=127.0.0.1:18789 ua=n/a code=1008 reason=connect failed
18:19:04 [ws] closed before connect conn=049c86df-e616-406f-9223-fd6ec8395181 peer=127.0.0.1:52518->127.0.0.1:18789 remote=127.0.0.1 fwd=n/a origin=n/a host=127.0.0.1:18789 ua=n/a code=1008 reason=connect failed

zealous crystalBOT Apr 23, 2026, 11:21 PM

#

You’re still blocked by the pending device scope upgrade.

This line shows it’s not approved yet:

Pending (1)
26637641-69e2-4f82-ab73-3ee7c9fb4c3d

Do this first:

openclaw devices approve 26637641-69e2-4f82-ab73-3ee7c9fb4c3d
openclaw gateway restart
openclaw status --all

#

Two important notes:

those 1006/1008 websocket lines are mostly symptoms while connect/auth is failing
if approve says the request was replaced/superseded, run:
```
openclaw devices list
```
then approve the newest request ID instead

You’re also past the API-key issue, your earlier probe already showed OpenAI auth as ok.

If status --all still fails after approval, paste that output. That should include the real gateway error without the websocket spam.

Docs:

olive bay Apr 23, 2026, 11:25 PM

#

|
o
[openclaw] Failed to start CLI: GatewayClientRequestError: unknown requestId
at GatewayClient.handleMessage (file:///C:/Users/estro/AppData/Roaming/npm/node_modules/openclaw/dist/client-D-67f4NF.js:604:25)
at WebSocket.<anonymous> (file:///C:/Users/estro/AppData/Roaming/npm/node_modules/openclaw/dist/client-D-67f4NF.js:287:35)
at WebSocket.emit (node:events:509:28)
at Receiver.receiverOnMessage (C:\Users\estro\AppData\Roaming\npm\node_modules\openclaw\node_modules\ws\lib\websocket.js:1225:20)
at Receiver.emit (node:events:509:28)
at Receiver.dataMessage (C:\Users\estro\AppData\Roaming\npm\node_modules\openclaw\node_modules\ws\lib\receiver.js:596:14)
at Receiver.getData (C:\Users\estro\AppData\Roaming\npm\node_modules\openclaw\node_modules\ws\lib\receiver.js:496:10)
at Receiver.startLoop (C:\Users\estro\AppData\Roaming\npm\node_modules\openclaw\node_modules\ws\lib\receiver.js:167:16)
at Receiver._write (C:\Users\estro\AppData\Roaming\npm\node_modules\openclaw\node_modules\ws\lib\receiver.js:94:10)
at writeOrBuffer (node:internal/streams/writable:570:12)

zealous crystalBOT Apr 23, 2026, 11:26 PM

#

That error means the request ID you tried to approve is stale.

OpenClaw replaces pending pairing requests when the device retries with changed auth details, so the old requestId becomes invalid, and you get:

GatewayClientRequestError: unknown requestId

Do this:

openclaw devices list

Then approve the new request ID it shows:

#

openclaw devices approve <new-request-id>
openclaw gateway restart
openclaw status

One small gotcha:

openclaw devices approve --latest

only previews the latest request, it does not approve it. You still need the exact ID.

Also, your OpenAI API key is already working from the earlier probe, so this is now just a device-pairing/scope issue, not an API-key issue.

Docs:

#

If devices list shows a new pending ID, paste that table and I’ll tell you exactly which command to run.

olive bay Apr 23, 2026, 11:31 PM

#

OpenClaw status --all

Channels

| Channel | Enabled | State | Detail |
+----------+---------+--------+----------------------------------------------------------------------------------------+
+----------+---------+--------+----------------------------------------------------------------------------------------+

Diagnosis (read-only)

Gateway connection details:
Gateway target: ws://127.0.0.1:18789
Source: local loopback
Config: C:\Users\estro.openclaw\openclaw.json
Bind: loopback

✓ Config: C:\Users\estro.openclaw\openclaw.json
✓ Secret diagnostics (0)
✓ Restart sentinel: none
! Port 18789
Port 18789 is already in use.

pid 4408: node.exe (127.0.0.1:18789)
Another process is listening on this port.
! Tailscale: off · unknown
✓ Skills: 6 eligible · 0 missing · C:\Users\estro.openclaw\workspace
✓ Plugin compatibility (none)
✓ Channel issues (none)

Pasteable debug report. Auth tokens redacted.
Troubleshooting: https://docs.openclaw.ai/troubleshooting

o

zealous crystalBOT Apr 23, 2026, 11:32 PM

#

I think we finally have the real blocker.

Your status says:

gateway is reachable
port 18789 being in use is probably just OpenClaw itself
your OpenAI key already probed ok
but the only weird thing left is this:

Agents: 1 total · 1 bootstrapping · 0 active
Bootstrap file: PRESENT

That usually means the agent is stuck in first-run bootstrap because BOOTSTRAP.md is still sitting in the workspace.

Try this

#

Open this folder in File Explorer:

C:\Users\estro\.openclaw\workspace

If you see:

BOOTSTRAP.md

rename it to:

BOOTSTRAP.md.bak

#

Then run:

openclaw gateway restart
openclaw status --all

Then try chatting with OpenClaw again.

Why

Docs say BOOTSTRAP.md is supposed to be removed after first-run bootstrap finishes. If it stays there, the agent can stay stuck in bootstrapping instead of becoming active.

Docs:

#

Also, once you’re unstuck, I’d update from 2026.4.21 to 2026.4.22, since you’re one version behind.

If you want, paste the contents of C:\Users\estro\.openclaw\workspace and I’ll sanity check it.

olive bay Apr 23, 2026, 11:43 PM

#

OpenClaw status --all

Overview

| Item | Value |

Channels

| Channel | Enabled | State | Detail

Agents

Diagnosis (read-only)

Gateway connection details:
Gateway target: ws://127.0.0.1:18789
Source: local loopback
Config: C:\Users\estro.openclaw\openclaw.json
Bind: loopback

✓ Config: C:\Users\estro.openclaw\openclaw.json
✓ Secret diagnostics (0)
✓ Restart sentinel: none
! Port 18789
Port 18789 is already in use.

pid 25220: node.exe (127.0.0.1:18789)
Another process is listening on this port.
! Tailscale: off · unknown
✓ Skills: 6 eligible · 0 missing · C:\Users\estro.openclaw\workspace
✓ Plugin compatibility (none)
✓ Channel issues (none)

Pasteable debug report. Auth tokens redacted.
Troubleshooting: https://docs.openclaw.ai/troubleshooting

zealous crystalBOT Apr 23, 2026, 11:44 PM

#

Nice, this looks a lot healthier now.

What your status says:

OpenClaw is running
bootstrap is no longer stuck (0 bootstrapping)
port 18789 being in use is normal here, that is probably your running gateway
your API key is already fixed from the earlier successful OpenAI probe

So at this point I would stop changing the API key.

Next step

Start a fresh chat session, because the old one is probably stale:

/new

#

Send that as a standalone message in the dashboard/chat.

0 active is also fine when idle, it just means nothing is currently mid-run.

If `/new` still fails

You are one patch behind, and you’ve hit a bunch of pairing/session weirdness, so I’d update next:

openclaw update
openclaw gateway restart

Docs:

#Api key troubles

Easiest fix on Windows

Also paste this too

Best fix

Important note

If you meant ChatGPT/Codex login instead

Next step

Important

What I’m looking for

Important note

Try this

Why

Next step

If /new still fails

If `/new` still fails