Pepedocs

Discord

Answer slash commands, or ordinary channel messages, in your Discord server with a Pepe agent.

Discord

Discord gives Pepe two different ways to receive a message, and a connection can use either or both:

Start with the guided setup (or the dashboard):

pepe setup

A connection’s config holds:

In the Discord app, point “Interactions Endpoint URL” at the connection URL and add a slash command with a text option (for example /ask prompt:...). Discord requires an acknowledgement within three seconds, so Pepe replies with a deferred response and posts the real answer as a follow-up once the agent finishes. Callback URL shape:

https://YOUR_HOST/webhooks/default/discord/<slug>

See Webhooks for the fields every connection shares (agent, mode, trainers, session_ttl_min, ephemeral, commands) and how the generic route works under the hood.

Files on a command

Give a slash command an attachment option and people can send Pepe a file with it: /ask prompt:what does he say? file:<clip>. A voice clip is transcribed before the agent runs, a document arrives with its text, and an image reaches a vision model as an image. The attachment is enough on its own, so /ask file:<clip> with nothing typed still works.

This is the only route a file has through the Interactions endpoint - it sees slash commands and nothing else, so a voice message or an attachment posted straight into a channel does not reach Pepe this way. It does reach Pepe the other way, over the gateway: see Ordinary channel messages below. Pepe accepts files up to 20 MB by default (max_attachment_mb on the connection can raise or lower that), and whichever of Pepe’s limit and Discord’s own upload limit is smaller wins: Discord’s own upload limit is 10 MB on a plain account or an unboosted server, higher with Nitro or a boosted server. See Voice messages and Documents.

Ordinary channel messages

A connection can also answer a message typed straight into a channel, a photo dropped into it, a voice message recorded in it, or a direct message to the bot - not just a slash command. This needs its own opt-in, because it means Pepe holds a persistent connection to Discord’s gateway rather than only answering webhook calls:

pepe gateway discord add support --agent helpdesk --gateway --bot-token '${DISCORD_BOT_TOKEN}'

or the equivalent fields in the dashboard (a connection can have both --application-id/--public-key for slash commands and --gateway/--bot-token for channel messages at once). mix pepe serve runs the connection either way. Flags:

Attachments of the message itself, of the message it replies to (so a voice note can be answered “what does this say?” after the fact), and of a message forwarded to the bot all count, and go through the same voice-to-transcript, document-to-text, image-to-vision handling as a slash command’s attachment. Messages on one connection are answered in the order Discord delivered them. If the connection drops, it reconnects and resumes automatically, with nothing said in the gap lost.

Switching models

The /model and /models commands let people check or change which AI model answers them. On Discord they reach Pepe through your registered command (/ask above): whatever you type in its prompt: option is the message Pepe sees. They work only on an admin-mode connection with commands enabled; on support, they are treated as plain text. /models lists the models available to this connection’s project; /model shows the current one, or changes it:

/model openrouter               # ask whether to switch just this chat or everyone
/model openrouter session       # switch for this conversation only
/model openrouter global        # switch for everyone this connection talks to

Anyone in an allowed conversation may switch the model for their own conversation. Switching it globally, for everyone this connection talks to, is reserved for trainers, the same trusted list that controls memory. Set model_switch_locked: true on the connection to turn model-switching off entirely for non-trainers.

← SlackMicrosoft Teams →