Discord Slash Commands ​

AIRI's Discord integration lets you interact with your AI directly from Discord using native slash commands. This page covers everything from getting started to the full command reference.

Setup: Getting a Bot Token ​

To use AIRI on Discord, you need a Discord Bot Token from the Discord Developer Portal.

Step-by-Step ​

1. Go to https://discord.com/developers/applications.
2. Click New Application, give it a name (e.g., "AIRI"), and create it.
3. Go to the Bot tab on the left sidebar.
4. Under the Token section, click Reset Token (or copy the existing one if available).
5. Copy the token — this is your DISCORD_BOT_TOKEN.
6. Under Privileged Gateway Intents, enable:
   • Message Content Intent (required for reading messages).
   • Server Members Intent (recommended).
   • Voice State Intent (required for voice commands like /summon).
7. Go to the OAuth2 > URL Generator tab.
8. Select scopes: bot and applications.commands.
9. Select permissions:
   • Send Messages
   • Read Message History
   • Use Slash Commands
   • Connect and Speak (for voice features)
   • Attach Files (for selfies and generated images)
10. Use the generated URL to invite the bot to your server.

Configure in AIRI ​

Once you have the token, open AIRI's Settings > Discord (under Messaging), paste the token, and enable the integration. AIRI will automatically register all slash commands with Discord.

How Slash Commands Work ​

AIRI's slash commands are registered with Discord's API using SlashCommandBuilder from discord.js. On service startup, AIRI pushes the command schema via Discord's REST API so your server sees the native / autocomplete UI.

When you type a command:

1. Discord sends an INTERACTION_CREATE event to AIRI.
2. AIRI's interaction handler maps the commandName to its local registry.
3. The corresponding execute function runs, showing "AIRI is thinking..." while processing.
4. The result is posted back to the channel as a native Discord message — often with rich embeds or file attachments.

All interactions flow through the same Episode/Memory system as the desktop app, so Discord conversations are synced with your chat history.

Tool Calls ​

AIRI's built-in tools (text journal, stickers, stage widgets, MCP, dating sim) are available from Discord — not just from the desktop Whisperbox. When you send a message from Discord, it enters the same LLM pipeline as desktop input. The tools array is resolved via a global fallback (toolsResolver), so you can ask AIRI to write into her journal, spawn a widget, or call MCP tools directly from a Discord channel.

The only case where tools are disabled is when you send an image attachment — vision models often don't support tool calling, and this applies to all sources equally (Discord, desktop, voice).

Command Reference ​

Core Commands ​

These are the essential commands available in the baseline integration.

Details ​

/character — Calls into the AIRI Card Store to load a different profile. The bot's identity (name, avatar, bio) syncs to Discord. All subsequent messages use the new character until you switch again.

/new — Clears short-term context and starts a fresh Episode. The optional msg immediately seeds the new conversation with a prompt.

/status — Queries main-process state stores and formats a rich embed with:

• Active character name
• Loaded LLM model
• Module state (vision, proactivity, etc.)
• Discord gateway health (ping/latency)

/history — Pulls recent messages from the active Episode's short-term memory and displays them as a formatted list.

/summon — Connects to the Discord Voice Gateway via @discordjs/voice. Audio routing depends on the active voice engine (see /voicecall).

/leave — Disconnects the active VoiceConnection and cleans up the audio pipeline.

Extended Commands ​

Advanced toggles and routing capabilities.

Details ​

/voicemode — Dictates how speech is delivered for text messages:

• puppet (default): Audio plays locally on the desktop app.
• voicenote: TTS chunks are combined into a single .ogg/.mp3 and uploaded as a voice note attachment.
• none: No TTS generated — saves API credits and resources.

/voicecall — Selects the real-time voice call engine:

• gemini: Discord Opus audio → Gemini Live WebSocket → Discord Opus. Ultra-low latency, no text serialization.
• classic: Audio → STT → LLM → TTS → Audio. Slower but uses high-fidelity desktop TTS providers and keeps a full text transcript.

/director — When on, AIRI can autonomously decide to generate images (widgets, backgrounds) through Artistry. When off, all generation must be explicitly requested.

/vision — When on, image attachments in Discord are intercepted, converted to base64, and fed into the VLM processor via the chat orchestrator. AIRI can "see" and respond to images.

/imagine — Sends the prompt to the configured Artistry backend (e.g., ComfyUI). The result is posted as a native Discord attachment.

/selfie — Triggers visionStore.heartbeat({ force: true }) — the same capture function used by the desktop Control Island. The screenshot is stored as a "selfie" entry in the Image Journal and posted as a high-quality attachment.

Telemetry & Settings UI ​

The Discord settings page at Settings > Discord (under Messaging) provides a real-time dashboard with:

Troubleshooting ​

Commands not showing up? Slash commands are registered on service startup. Try restarting the Discord integration from Settings, or re-invite the bot with applications.commands scope.

Bot not responding? Check the token is valid and the bot has the required intents enabled (Message Content, Voice State) in the Discord Developer Portal.

Voice not working? Ensure the bot has Connect and Speak permissions in the voice channel, and that @discordjs/voice dependencies are installed.

Further Reading ​

• Discord Developer Portal — Bot Tokens
• discord.js Guide — Slash Commands
• Feature Showcase: Discord Integration