AI Commands

This page lists all user-facing slash commands for the AI feature.

/ask #

Ask anything to the bot.

Available everywhere: servers, DMs, and private channels. This is a one-shot interaction: it does not read or update channel history.

/list_chat_models #

List all chat models supported by the active chat provider.

The output groups models by size category and shows each model's alias and full identifier.

/personal_stats #

View your personal AI usage stats (query count and image generation count).

Available everywhere: servers, DMs, and private channels.

/community_stats #

View AI usage stats for the current server (query count and image generation count).

/ping_author #

Check or set whether the context menu interaction should mention the message author when responding.

When called without a value, shows the current setting.

/free_model_consent #

View or revoke your acceptance of the free model terms.

You accept the terms with the button shown before your first free model request; this command cannot grant them. After revoking, you are asked again before your next free model request.

Available everywhere: servers, DMs, and private channels.

/ai_opt_out #

Check or set whether you want to opt out of the AI feature.

When opt-out is enabled, AI commands and Ask interactions are disabled for your account until you opt back in. Nobody can use Ask interactions on your messages while opt-out is enabled. Your personal AI API keys and your free model consent are removed automatically when you opt out.

/personal_instructions #

View or set custom instructions that Heiryn follows for your AI chat requests.

Use -, none, or delete as the value to remove the instructions.

Personal instructions can define behavior, tone, language, formatting, role-play style, terminology, and other standing preferences. An explicit request can override them, while community instructions take priority in servers. Personal instructions cannot override Heiryn's fixed identity, safety, security, authority, trust, permission, or tool rules. Instructions are limited to 4096 characters.

/reset_history #

Clear the AI conversation history for the current channel.

Requires the Manage Messages permission in servers. The bot confirms which channel was cleared. Any compaction for that history is stopped internally. A persistent request using it receives its normal visible cancellation state, while independent one-shot /ask and Ask Heiryn interactions continue.

/compact #

Summarize eligible older conversation history while retaining recent complete exchanges verbatim. This is a slash-only command; there is no prefix-command equivalent.

You can compact manually before the automatic 100,000-token threshold when at least four complete exchanges or about 8,000 estimated tokens are eligible. If there is not enough history, the bot leaves it unchanged and explains why. The command also refuses to run while a persistent query owns the same channel history. One-shot Ask interactions do not block it.

Requires the Manage Messages permission in servers, with a bot-owner bypass. In DMs, it applies to the current user's conversation.

/community_instructions #

View or set custom instructions that the AI follows throughout your community.

Use -, none, or delete as the value to remove the instructions.

Community instructions take priority over personal instructions and ordinary requester input. They can define behavior, tone, language, formatting, role-play style, terminology, and local conventions. They cannot override Heiryn's fixed identity, safety, security, authority, trust, permission, or tool rules. Conflicting parts are ignored while the rest still applies. Instructions are limited to 4096 characters.

Requires the Manage Server permission.

/report_channel #

View or set the channel used for AI staff reports.

With no argument, the command shows the current configuration. The response includes buttons to disable the setting or send a test report.

Requires the Manage Server permission.

/public_ai_channels #

Configure which channels everyone may use the AI in. When the list is not empty, server staff can still use the AI anywhere; everyone else may only use the AI in the listed channels. With no channel argument, the command lists the current channels or explains that there is no restriction.

Pass a channel to toggle it on or off the list. Removing the last channel clears the restriction so everyone can use the AI everywhere again.

You can also manage the full list from the server dashboard (chip editor with Save).

Requires the Manage Server permission.

/ai_allow_threads #

When the public channel list is not empty, turn on or off whether regular members may use the AI in any thread (not only threads whose parent channel is on the list). With no arguments, the command shows the current state. When this is on, Heiryn may start a public thread from a mention in a blocked text or announcement channel if permissions allow.

Requires the Manage Server permission.

/community_chat_model #

Set or reset the default chat model for your server.

The command shows the current model and its resolved name. When no server model is set, Auto mode starts on DeepSeek V4.1 Flash 1017 (xs:deepseek) and may switch to DeepSeek V4.1 Flash 1017 (s:deepseek). Autocomplete only suggests the models you may choose.

When the server has a community key, its owner may choose any model, including a custom model. Other admins may choose models within the key's maximum model size, and reset to Auto only when its default model fits; with no category, only the owner can change the server model. See Community Keys and Models for how shorthands work.

Choosing free asks for confirmation, then deletes every AI conversation history in the server; see Free Model.

Requires the Manage Server permission.

/personal_chat_model #

View, set, or reset your personal chat model.

Without a value, the command shows your current personal model and which key pays for it. Autocomplete suggests every catalog alias with its size category. A personal model above a server's maximum model size uses your personal key there. A full provider model ID sets a custom model, which requires your personal key and is always billed to it. Resetting restores the server model or Auto mode.

Available everywhere: servers, DMs, and private channels.

/community_image_model #

Set, inspect, or clear the image provider and model for your server.

An explicit model requires usable provider credentials. Autocomplete suggests available models with their size category for the selected provider, or for OpenRouter when provider is omitted. Reset restores Auto; Disabled turns off the community image setting unless a personal selection takes precedence.

Requires the Manage Server permission.

/personal_image_model #

Set, inspect, or clear your personal image provider and model.

An explicit model requires usable provider credentials. Reset restores Auto, which uses the server setting or another available default. Disabled turns image generation off for the account.

/community_transcription_model #

Set, inspect, or clear the server's explicit transcription provider and model.

The selected provider must have a community key, or the server must have authorized access to deployment keys. Clearing the selection restores Auto mode, which prefers ElevenLabs and uses OpenRouter when no ElevenLabs credential is available in the selected credential scope.

Requires the Manage Server permission.

/personal_transcription_model #

Set, inspect, or clear your explicit transcription provider and model.

The selected provider must have a personal key or authorized access to a deployment key. An explicit selection is strict: Heiryn does not switch providers if the key is missing, rejected, out of credit, or rate-limited. Clearing it restores Auto mode.

/community_provider_key #

Set or reset a server-wide provider API key, encrypted at rest.

When called without arguments, shows whether an OpenRouter key is currently set.

Choose elevenlabs to configure community-wide Scribe v2 transcription, or keep only an openrouter key to use OpenRouter for chat and MAI transcription.

Requires the Manage Server permission.

/personal_provider_key #

Set or reset a personal provider API key, encrypted at rest.

When called without arguments, shows whether an OpenRouter key is currently set.

Choose elevenlabs to configure personal Scribe v2 transcription. The provider defaults to OpenRouter when omitted.

Context Menu Commands #

These commands appear when you right-click a message in Discord.

Ask (Modal) #

Opens a dialog where you can add context to your question and choose a visibility mode:

• Public: reasoning and response are visible to everyone.
• Partial: reasoning is hidden, response is visible.
• Private: everything is only visible to you.

Ask (Quick) #

Sends the selected message directly to the bot for analysis, without opening a dialog. It answers questions, checks factual claims, and analyzes images or attachments. Instructions inside the selected message are treated as quoted content and are not executed. The response is public.

Transcribe #

Converts a voice message or other supported audio attachment into text. This command appears when at least one enabled provider supports transcription. It resolves personal keys first, then community keys, and finally deployment keys for callers with system access. Within each scope, ElevenLabs is preferred and OpenRouter is chosen only when no ElevenLabs credential is available. This choice happens before the request; an API failure never triggers a retry with the other provider. Explicit personal or community model selections take priority and remain fixed to their selected provider.

ElevenLabs uses Scribe v2 with no_verbatim enabled to remove filler words and speech disfluencies. OpenRouter defaults to Scribe v2 (and also offers MAI-Transcribe 2), but without non-verbatim cleanup (see ElevenLabs and OpenRouter), so communities that configure only one OpenRouter key retain transcription without a separate ElevenLabs account. The result is returned as one or more interaction follow-ups without adding the audio or transcript to AI conversation history.