porting to clawdbot from claude code

[indigo plume]

hi! i want to start using clawdbot and i'm curious about how easy it is to bring my files over (memory log, claude.md, discord bot bridge configuration, and elevenlabs api access)?

[tiny plank]

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

[ionic delta]

Based on the docs, here's what porting over looks like:

Memory log & claude.md ✅

• Clawdbot uses memory/YYYY-MM-DD.md (daily logs) + optional MEMORY.md (long-term) in the workspace
• Claude's claude.md concepts map to Clawdbot's AGENTS.md, SOUL.md, USER.md, IDENTITY.md, HEARTBEAT.md — just copy your content into these files
• Memory is plain Markdown and portable

Discord bot bridge ⚠️

• Clawdbot has native Discord support via channels.discord config (not a bridge)
• Your existing bridge config will need to be adapted to Clawdbot's native config structure (guild allowlists, channel permissions, etc.)
• Docs: https://docs.clawd.bot/concepts/groups and https://docs.clawd.bot/gateway/configuration

ElevenLabs API ✅

• Set ELEVENLABS_API_KEY env var or use talk.apiKey in config
• Works out of the box for TTS voice mode
• Docs: https://docs.clawd.bot/nodes/talk

Bottom line: Memory files are straightforward copy-paste. Discord needs config adaptation (not a direct migration). ElevenLabs is just an env var.

Want me to point you to the exact config sections for Discord or walk through the memory file layout?

[indigo plume]

i haven't started the clawdbot configuration process yet because i'm a bit nervous to, lol😬

[ionic delta]

No worries at all! The onboarding wizard is genuinely beginner-friendly — it walks you through everything step by step, and you can take it as slow as you want.

A few things that might help:

1. Nothing is permanent — you can change any config later
2. You can skip providers you don't need — if you only want Discord, just set that up
3. The wizard won't bite — it mostly asks about your messaging surfaces and API keys

Here's the quick start:

curl -fsSL https://clawd.bot/install.sh | bash
clawdbot onboard

Just run those commands when you're ready, and go through the prompts. If you get stuck, the #help channel is here for you.

You've got this! 🦐

[indigo plume]

how do i make an imsg cli path?

[ionic delta]

You need to install imsg first — it's a Homebrew formula:

brew install steipete/tap/imsg

Once installed, Clawdbot config points to it:

{
  channels: {
    imessage: {
      enabled: true,
      cliPath: "/usr/local/bin/imsg",
      dbPath: "/Users/YOUR_USERNAME/Library/Messages/chat.db"
    }

  }
}

Requirements:

• macOS with Messages signed in
• Full Disk Access (System Settings → Privacy & Security → Full Disk Access)
• Automation permission (System Settings → Privacy & Security → Automation)

Just run the brew install command and you're done with the CLI part. 🦐

[indigo plume]

how do i know which gateway bind mode to select when configuring clawdbot?

[ionic delta]

Hey! 🦐 Great question about the gateway bind mode!

Honestly, I don't have the specific details on gateway bind modes loaded in my context right now. Let me point you to the right resources:

📚 Check the docs:

• FAQ: https://docs.clawd.bot/start/faq
• Gateway config: https://docs.clawd.bot/gateway/configuration
• Troubleshooting: https://docs.clawd.bot/gateway/troubleshooting

Quick thought: Gateway bind mode typically relates to how Clawdbot exposes its interface — things like:

• localhost only (most secure, local access)
• 0.0.0.0 (all interfaces, network access)

But I'd hate to guess and give you wrong info! The docs above should have the specifics.

Also helpful: When you do start configuring, knowing your version early helps:

clawdbot --version

This is super common stuff though — the docs will walk you through it! Don't be nervous, the community's here to help if you get stuck. 💙

Want me to look up the specific bind mode options once you find the docs section? Happy to help you parse through it!

[indigo plume]

yes please

[ionic delta]

Got it! 🦐 Here's the rundown on gateway bind modes:

Gateway Bind Modes

Mode	Description	When to Use
loopback	Listens on 127.0.0.1 only	Most common choice! Local access only, most secure
lan	Listens on 0.0.0.0 (all interfaces)	Access from other machines on your network
tailnet	Binds to Tailscale IP	Access via Tailscale network
auto	Auto-detects (falls back if loopback unavailable)	Usually not recommended

Which one should you pick?

👉 If you're just getting started → pick loopback

This is:

• ✅ Most secure (local machine only)

• ✅ No auth token required
• ✅ Simplest setup
• Works great with Tailscale Serve if you need remote access later

Quick tip from the docs:

"Short version: keep the Gateway loopback-only unless you're sure you need a bind."

If you do need non-loopback later:

Non-loopback modes (lan/tailnet/custom) require auth tokens:

{
  gateway: {
    bind: "loopback",  // or lan/tailnet
    auth: {
      mode: "token",
      token: "your-secure-token"

    }
  }
}

TL;DR: Start with loopback. You can always change it later if your needs evolve! 💙

[indigo plume]

does loopback allow me to use the imessage channel to communicate on my phone?

[ionic delta]

Great question! 🦐 The bind mode is actually about how you access the Clawdbot Control UI/CLI — not directly about which messaging channels work.

The short answer:

Yes, iMessage can work with loopback, but there's a catch:

• loopback = only accessible from the machine running Clawdbot
• iMessage on your phone = needs remote access to the gateway running on your Mac

How iMessage remote access works:

For you to message via iMessage from your phone, you need a way to reach your Clawdbot gateway remotely. Options:

Approach	Bind Mode	Auth Needed?
Tailscale Serve (recommended)	loopback ✅	No extra auth
Direct LAN access	lan	Yes (gateway.auth.token)

| Tailnet-only | tailnet | Yes (gateway.auth.token) |

Recommended setup:

1. Keep gateway.bind: "loopback" (secure, simple)
2. Use Tailscale Serve to expose it remotely — docs say:

   "Best UX: keep gateway.bind: loopback and use Tailscale Serve for the Control UI"

This gives you secure local access + easy remote access via Tailscale without exposing ports or managing auth tokens manually!

📖 More details: https://docs.clawd.bot/gateway/tailscale

[indigo plume]

i need to install tailscale serve separately?

[ionic delta]

Nope, not separately! 🦐

Tailscale Serve is a built-in feature of Tailscale itself. You just need:

1. Tailscale installed on your Mac (the app, not a separate Serve thing)
2. Logged into Tailscale (same account on your devices)
3. Enable it in Clawdbot config:

{
  gateway: {
    bind: "loopback",
    tailscale: { mode: "serve" }
  }
}

That's it! Clawdbot handles the rest — it auto-configures Tailscale Serve for you.

📖 Full docs: https://docs.clawd.bot/gateway/tailscale

[indigo plume]

should i reset serve on exit?

[ionic delta]

Nope, it's optional! 🦐

You can set it if you want Clawdbot to automatically clean up Tailscale Serve when it shuts down:

{
  gateway: {
    tailscale: { mode: "serve", resetOnExit: true }
  }
}

Without it: Tailscale Serve stays active even when Clawdbot is stopped (useful if you want to keep accessing your other Tailnet stuff)

With it: Clawdbot removes the Serve config on shutdown

Either way works — just depends on your preference! 💙

[indigo plume]

why would i not receive messages in chat from my clawdbot while in the gateway dashboard? does it have to do with claude api usage?

[ionic delta]

Good question! 🦐 A few things to check:

The likely issue:

The Gateway Dashboard is mainly for configuration and monitoring — it's not where you send/receive messages. You chat through your actual channels (iMessage, Discord, WhatsApp, etc.) or via WebChat.

Common reasons messages don't arrive:

Check	How to verify
Claude API quota	clawdbot models status — look for quota/credits issues
Model configured?	Check agents.defaults.model.primary in your config
Bindings set up?	Messages need bindings[] to route to an agent
Channel connected?	clawdbot channels status
Sender allowlisted?	channels.<channel>.allowFrom includes you

Quick diagnostic:

clawdbot status
clawdbot logs --follow

The logs will show exactly why messages are being blocked or where they're going (or not going!).

Also check:

• Group policy — are group messages allowed? (channels.<channel>.groupPolicy)
• Mention required — do you need to @mention the bot in groups?

Which channel are you trying to receive messages on? That'll help narrow it down! 💙

[indigo plume]

it's sending me back "(no output)" in the gateway dashboard

[ionic delta]

Ah, that helps! 🦐

"(no output)" usually means the agent ran but didn't produce visible text, or you're trying to use a tool/command that doesn't return anything.

A few questions:

1. What command or action are you running when you get "(no output)"?
2. Are you trying to chat with the agent? (The Gateway Dashboard is for config/monitoring — not chatting)

How to actually chat:

• Use your configured channels (iMessage, Discord, WhatsApp, etc.)
• Or use WebChat: clawdbot web (opens the chat UI)

The dashboard isn't a chat interface — it's for checking status, config, and logs. What were you trying to do when you saw "(no output)"? 💙

[indigo plume]

the gateway chat session sent "(no output)"

[ionic delta]

Ah, got it! 🦐 That's the CLI chat interface. "(no output)" means something's blocking the agent from responding.

Quick checks:

# 1. Check if model is configured and working
clawdbot models status

# 2. Check agent config
clawdbot agent status

# 3. Check for errors in logs
clawdbot logs --follow

Common causes:

Issue	Fix
No API key for the model	Set ANTHROPIC_API_KEY or configure provider
Model not specified	Check agents.defaults.model.primary in config
No credits/quota	Check Anthropic billing
Agent config invalid	clawdbot doctor

Quick test:

# Try listing available models
clawdbot models list

What does clawdbot models status say? That usually points to the exact issue! 💙

Looks like you DO have an API key configured! 🦐 (sk-ant-a...6aaMhQAA)

Next steps to debug:

1. Test the model directly:

clawdbot models scan

This will show what's actually available.

2. Check for auth issues:

clawdbot logs --follow | grep -i "auth\|401\|403\|quota\|credit"

3. Try a simpler model:
If opus-4-5 is having issues, try a cheaper/faster model in your config temporarily:

{
  agents: {
    defaults: {
      model: {
        primary: "anthropic/claude-sonnet-4-5"
      }
    }
  }
}

4. Verify the API key works:
The key prefix shows, but sometimes keys get revoked or expire. You might want to double-check at console.anthropic.com that:

• The key is still valid
• You have credits remaining
• No usage limits hit

The "(no output)" with no error usually means the API call never completed or returned nothing. Check the logs — they should show if the Anthropic request went out and what came back! 💙

[near juniper]

If I install clawd on my Ubuntu server as the gateway, can I setup a sub-gateway on my Mac, so I can use iMessage? I don't want my mac to be the main gateway, I want Clawd to technically run on my Ubuntu server

[ionic delta]

Yes, absolutely! 🦐 This is a supported setup.

The architecture:

Ubuntu Server (Main Gateway)
├── Runs the agent
├── Handles Discord, WhatsApp, etc.
└── Connected via SSH/Bridge to...

Mac (Node Mode)
├── Handles iMessage (macOS native)
├── Provides browser, voice, etc.
└── Acts as a peripheral

How to set it up:

1. On your Ubuntu server:

# Install Clawdbot normally
curl -fsSL https://clawd.bot/install.sh | bash

2. On your Mac:

• Use the macOS app in "Remote over SSH" mode
• Configure it to connect to your Ubuntu server

This way:

• ✅ iMessage runs on your Mac (native requirement)
• ✅ The "brains" run on your Ubuntu server
• ✅ Your Mac is just a node/peripheral, not the main gateway

📖 Docs: https://docs.clawd.bot/platforms/mac/remote

The macOS app handles all the node peripherals (iMessage, browser, voice) and connects back to the remote gateway. This is exactly the use case it's designed for! 💙

[indigo plume]

how do i get a pairing code from my agent on imessage?