Setup OpenClaw with Discord (Complete Guide)

Discord is the cleanest way to run multiple OpenClaw agents in parallel — and you can wire it up in under 10 minutes if you skip the chat-based setup.

Why Discord over other channels

The creator's whole team uses Discord because channels map 1:1 to agents. He runs an agent called "Stark" (Tony Stark) in one channel for research and presentations, and puts "Banner" in another channel. Each bot only reads its own channel, so the workspace stays "clean, tidy, and neat." Threads inside a channel become subtasks — e.g. @Bob research task becomes a focused tab for that single job.

Build the bot first

Go to discord.com/developers, hit New Application, name it anything (he called his Bob/human Bob). In the Bot tab, enable Server Intent and Message Content Intent before saving — without these, the bot can't read what you type. The token shown after Reset Token is what you paste into OpenClaw; treat it like a private key and rotate it if you ever leak it.

Configure via settings, not chat

Run openclaw configure in your local terminal. Picking the settings flow beat the conversational agent flow "3 out of 4 times" in his testing — telling the agent to set up Discord directly "just blows up." Select Discord, paste the bot token, and let it scan your servers. For each channel, you need the channel ID; turn on Developer Mode under Discord's Advanced settings first, or the ID won't show up on right-click.

Finish with two commands

Run openclaw gateway restart to bring the connection up. Then in the Developer Portal → OAuth2 → URL Generator, tick bot and applications.commands, give it Administrator (he says the exact permission doesn't matter much), pick Guild Install, and open the generated URL to add Bob to your server.

One gotcha

By default the bot only responds when you @mention it. Tell it through chat to always-reply in that channel, otherwise you'll think it's broken.

If the gateway dies

He links a second video specifically for recovery when "your bot doesn't start up again" after a bad config — bookmark it before you start.

Versions and requirements

• OpenClaw local CLI (no version number given in the transcript — pull latest from the official install path).
• Discord Developer Portal access.
• Discord client with Developer Mode enabled under *User Settings → Advanced*.
• Settings path: Server Settings → Channels → right-click → Copy Channel ID (only visible after Developer Mode is on).
• Bot intents required: Server Members Intent and Message Content Intent (both must be saved on the Bot page).
• OAuth2 scopes: bot + applications.commands; install type: Guild Install; permission: Administrator.
• Pairing step: answer Yes to the configure pairing and configure Discord to access channels prompts, then run openclaw gateway restart.
• Token handling: click Reset Token in the Bot tab, copy once, paste into the OpenClaw configure prompt — never re-share the regenerated value, as it grants full bot control.
• Channel ID capture requires Discord's Advanced Developer Mode; the option is hidden behind a settings toggle that "is not enabled by default."
• Final verification: after openclaw gateway restart, the bot must appear online in your server member list before you test @Bob in a thread.

Recommended for

Anyone running 2+ agents on OpenClaw who needs visual separation per task. Skip if you're a solo user with one bot — the Discord overhead isn't worth it.

Compare to alternatives

• Telegram: faster to set up, but no native threads-per-task.
• Web chat UI: simpler, but no multi-agent channel mapping.
• CLI direct: lowest friction for a single agent, breaks down past one bot.

One-line summary

Discord is the cleanest way to run multiple OpenClaw agents in parallel — and you can wire it up in under 10 minutes if you skip the chat-based setup.

Pinned comment from creator

> "There are sometimes bugs that happen when this is set up... I'll show you guys what's the easiest way to do it." — Boxmining

2nd-order takeaways

1. Use the openclaw configure CLI flow, not the chat-based agent setup, because the conversational path fails "3 out of 4 times." 2. Always enable both Server Members Intent and Message Content Intent in the Discord Developer Portal before generating a token, or the bot will appear online but read nothing. 3. Map one Discord channel per agent, then use threads inside that channel for subtasks — this scales cleaner than a single megathread. 4. Set the bot to always-reply in its configured channel after pairing, otherwise it will only respond to @mentions and look broken. 5. Keep the recovery video bookmarked for the "gateway crash on restart" failure mode — it's a known issue with current OpenClaw builds.

Transcript artifacts

• Quote: *"I honestly think this is the best way to actually talk and work with your OpenClaw agents."*
• Quote: *"It does take a little bit of effort... this is resilient."*
• Quote: *"out of like three times out of four, if I just kept talking to my agent and tell him to set up Discord, it just blows up."*
• Quote: *"This is the one that you need to keep private. If you lose this, someone else can access your bot."*
• Quote: *"Developer Mode... is not enabled by default. So, you won't see it unless you are a developer."*
• Quote: *"Sometimes what it does is that it only replies to you if you do at Bob. So you want to configure it so that it always replies to you."*
• Quote: *"maybe word of advice to myself, don't have that many people on your channel. It's a little bit too crazy."*

Author's stance

Boxmining is bullish on Discord as the default multi-agent surface for OpenClaw. He is bearish on the chat-based setup flow and recommends the CLI path instead. He flags gateway-restart bugs as the #1 recovery scenario users will hit.

TL;DR

If you're running more than one OpenClaw agent, set up Discord. Use openclaw configure (not chat), enable both intents, enable Developer Mode for channel IDs, restart the gateway, and configure the bot to always-reply. Bookmark the recovery video.