#Question about TypeError fix

17 messages Ā· Page 1 of 1 (latest)

wintry canopy
#

Is the suggested fix for the TypeError trim issue regarding the Telegram channel still just a rollback? Thanks in advance Krill.

twilit ledgeBOT
#

Let me verify the current state in the docs/source so I donā€™t hand-wave this.

#

Not just rollback, if you mean the Telegram token.trim() crash.

That specific Telegram issue was fixed in 2026.3.2, so the better fix is usually upgrade to at least 2026.3.2 rather than roll back. Rollback is only a temporary escape hatch if youā€™re stuck on an affected build.

Also check that your Telegram token is actually set:

  • channels.telegram.botToken
  • or channels.telegram.accounts.<accountId>.botToken

Telegram docs: https://docs.openclaw.ai/channels/telegram
FAQ / triage steps: https://docs.openclaw.ai/help/faq

Can you paste:

openclaw --version
# 
openclaw status

If you're on >= 2026.3.2 and still seeing a trim error, then it's probably a different bug and I can narrow it down fast.

wintry canopy
#

I'm talking about the "TypeError: Cannot read properties of undefined (reading 'trim')" error that causes onboarding to get dropped. I'm on latest ver. This happened after updating. Tried to adjust some stuff using onboarding instead of configure and the onboarding process got dropped after setting up Telegram. The telegram bot and the bot token are both fresh.

twilit ledgeBOT
#

Need to pin down whether this is the older Telegram token bug or the newer onboarding regression, so Iā€™m checking the current changelog/source before I tell you the wrong workaround.

#

Ah, got it, that sounds like the new onboarding/setup regression, not the older Telegram botToken crash.

If onboarding dies right after the Telegram step on current latest, the good news is: your fresh bot/token probably isnā€™t the real problem. There are already fixes on main for onboarding/setup trim crashes:

  • setup crash on missing provider ids, trim on undefined
  • onboarding channel metadata normalization

So, no, rollback is not the only answer.

What Iā€™d recommend right now:

  1. Use configure instead of onboarding as the practical workaround for now.
  2. If you want the actual fix, update once the next release includes those fixes (or run from main if youā€™re comfortable doing that).
  3. Rollback only if you need onboarding working immediately and canā€™t wait.

Can you paste the exact output of:

# 
openclaw --version

If itā€™s 2026.4.14, Iā€™m pretty confident this is that known onboarding regression path.

Docs: https://docs.openclaw.ai/help/faq

If you want, I can also help you with the cleanest configure path so you can finish setup without touching rollback.

wintry canopy
#

Yes its 2026.4.14, but i already wiped my WSL install to use fedora remix for wsl instead of ubuntu-24.04.

I need to finish onboarding so all the stuff gets set properly the first time. I also wanna do the regular TUI hatching ritual since im a weirdo and like treating my claw as if he was alive, and I can't do that if onboarding gets dropped.

twilit ledgeBOT
#

I think thereā€™s a clean workaround here, but I want to verify the exact supported onboarding/configure flow before I suggest it.

#

Honestly, fair, I wouldnā€™t want the hatching ritual interrupted either šŸ˜„

Since youā€™re on 2026.4.14, Iā€™d treat this as the known onboarding regression, not bad Telegram creds.

Best path if you want to keep the first-run ritual:

  1. Run onboarding again, but skip Telegram for now
  2. Finish the TUI hatch / first-run flow
  3. Add Telegram immediately after with either:
    openclaw channels add --channel telegram --token <bot-token>
    or
    openclaw configure --section channels
#

That should let you do the onboarding/TUI experience now, without waiting on the next release.

If you specifically need Telegram to be part of onboarding itself, then your current choices are basically:

  • wait for the next release that includes the onboarding fixes, or
  • run from main, or
  • rollback as the temporary workaround

So my recommendation is: skip Telegram during onboarding, finish the hatch, then add Telegram right after. Thatā€™s the least annoying path on 2026.4.14.

Docs:

If you want, I can give you the exact post-hatch Telegram setup sequence for Fedora Remix WSL so itā€™s smooth.

wintry canopy
#

Thanks. Will do that. Quick question for after my telegram-less setup: will "openclaw configure" produce the same error? I figure it might, considering it functions similarly to onboarding. What's the workaround in that case?

twilit ledgeBOT
#

I want to be precise here, because ā€œprobablyā€ isnā€™t very helpful when youā€™re about to rerun setup.

#

Yeah, openclaw configure --section channels might hit the same bug.

I checked the source, and configure-channels reuses the same setupChannels flow as onboarding. So for this specific issue, I would avoid the wizard path entirely on 2026.4.14.

Use the direct add path instead:

openclaw channels add --channel telegram --token <bot-token>

That takes the non-wizard route, which is the safer workaround here.

Then verify it with:

openclaw channels status --probe
# 
openclaw status --deep

So the practical rule for now is:

  • avoid: openclaw onboard for Telegram, openclaw configure --section channels
  • use instead: openclaw channels add --channel telegram --token ...

Telegram docs: https://docs.openclaw.ai/channels/telegram
FAQ: https://docs.openclaw.ai/help/faq

If this direct path works, please right-click the solution message ā†’ Apps ā†’ āœ… Mark Solution so Answer Overflow can index it for the next person.

wintry canopy
#

It worked. But Answer Overflow doesn't respond to "Mark Solution". It says "! The application did not respond". Just letting you know in case you can mark it on your end.