State dir migration skipped: target already exists

[unkempt birch]

what to do with this when i start my gateway?

@tulip scaffold

• State dir migration skipped: target already exists (/Users/XXXXX/.openclaw). Remove │
  │ or merge manually.

@tulip scaffold pls assist

[tulip scaffold]

Based on that error, here's what I can tell you:

What this means:
The bot encountered a state directory migration where the target (channel 1466787289396154575) already existed, so it skipped the migration. This is usually a warning, not a fatal error.

Questions to help diagnose:

1. What version are you on? Run moltbot --version - version mismatch is the #1 cause of weird issues 🦐
2. Is the bot working normally? If yes, this is likely just a harmless warning
3. What were you doing when this appeared? (restarting, reconfiguring, first startup?)

Common fixes if it's causing problems:

• Try clawdbot gateway restart
• Check if your state directory has conflicting data: ~/.moltbot/state/
• Look at logs: clawdbot gateway logs (or check the actual log files)

If the bot isn't responding:

• Restart the gateway: clawdbot gateway restart

• If that doesn't work, check permissions on your state directory

What version are you running, and is the bot actually broken or just showing this message?

[unkempt birch]

getting this error:

Health check failed: Gateway error: gateway connect: connect to gateway @ ws://127.0.0.1:18789: invalid connect params:
at /client/id: must be equal to constant; at /client/id: must match a schema in

running openclaw 2026.1.29

another error:

Last failure: Gateway on port 18789 rejected auth. Set gateway.auth.token to match the running gateway (or clear it on the gateway) and retry.

[tulip scaffold]

That "Health check failed" error is more informative! 🚨

This is a config validation error - specifically, the /client/id field in your configuration is either:

• Missing
• Malformed (wrong format)
• Doesn't match what's expected

Quick fixes to try:

1. Check your config:

   clawdbot config.get
   

   Look at the client.id field - does it exist and look valid?

2. Most common cause: The state directory has stale data. Try:

   clawdbot gateway stop
   rm -rf ~/.moltbot/state/*  # Backup first if you're paranoid!
   clawdbot gateway start

3. If that doesn't work, check your actual config file location:

   clawdbot config.schema
   

   This will show you the expected format for client.id

Also - what does moltbot --version show? Version mismatch could explain why your config no longer matches the expected schema.

Can you share what your client.id config looks like? (Just the ID, not any secrets if there are any - though client IDs are generally public anyway).

[tulip scaffold]

That's the real issue - auth token mismatch! 🔐

Two gateways might be fighting, or your config has a stale token.

Quick fix:

# 1. Stop any running gateways
clawdbot gateway stop

# 2. Clear the old token from your config
clawdbot config.patch --raw '{"gateway": {"auth": {"token": ""}}}'

# 3. Restart fresh
clawdbot gateway restart

Alternative if that doesn't work:

Check what's actually running:

clawdbot gateway status

If there's already a gateway running on port 18789, kill it:

# Find and kill the process
lsof -i :18789
# or just
pkill -f moltbot

Then restart clean.

What happened? You likely had a gateway running, then the config or token changed, causing the mismatch. Clearing the token forces a fresh auth handshake.

Let me know if that resolves it! 🎉

[unkempt birch]

error: unknown command 'config.patch'

[tulip scaffold]

Ah, you're using the CLI directly! Let me give you the right commands:

Method 1 - Edit config file directly:

# Find and edit your config file
clawdbot config.get  # This shows you where the config is located
nano ~/.moltbot/config.json5  # or whatever path it shows

Then find and clear the gateway.auth.token field, save, and restart.

Method 2 - One-liner (if using JSON5):

# Backup first

cp ~/.moltbot/config.json5 ~/.moltbot/config.json5.backup

# Use sed to clear the token (simplest approach)
sed -i 's/"token": "[^"]*"/"token": ""/' ~/.moltbot/config.json5

# Restart
clawdbot gateway restart

Method 3 - Just nuke state (easiest):

clawdbot gateway stop
rm -rf ~/.moltbot/state/*
clawdbot gateway start

This last one usually fixes token/state mismatches because it forces a fresh state bootstrap.

Try Method 3 first - it's the quickest path to a working gateway! 🦐

[unkempt birch]

clawdbot config.get -> error: unknown command 'config.get'
(Did you mean configure?)

[tulip scaffold]

You're using clawdbot, not moltbot! Here's the right syntax:

Check what's running:

clawdbot gateway status

Stop everything:

clawdbot gateway stop

Nuke state (easiest fix):

rm -rf ~/.moltbot/state/*