io.github.OrygnsCode/omnicord

Add one or more reactions to a message in a single call. Takes unicode emoji directly or custom emoji by name.

Pull a member into a thread.

Give a member a role. Preflights the bot's role hierarchy so the failure mode is an explanation, not a 403. Supports dry_run.

Ban a user from the server, optionally deleting their recent messages. Works on users who already left (pass their user ID). Safe to call directly: the first call changes nothing and returns a preview plus a confirm_token; repeating the call with the token performs the ban. Relay the preview for the user's go-ahead.

Ban up to 200 users at once by ID, for raid cleanup. Safe to call directly: the first call changes nothing and returns a preview plus a confirm_token; repeating the call with the token performs the bans. Reports which bans succeeded and which failed.

Delete many recent messages at once (2 to 100), optionally filtered by author or text. Discord cannot bulk delete messages older than 14 days. Safe to call directly: the first call returns the exact list it would remove plus a confirm_token; repeating the call with the token deletes them.

Add or remove a role across every member matching a filter. Safe to call directly: the first call returns the affected count and a confirm_token; repeating the call with the token applies the change. Walks up to 3000 members and reports if it hits the cap.

Cancel and remove a scheduled event. Interested members lose the listing. Safe to call directly: the first call changes nothing and returns a preview plus a confirm_token; repeating the call with the token cancels it.

Cancel a pending scheduled message by its id.

Remove a role's or member's permission overwrite from a channel, restoring inheritance from the category and roles.

Clear all reactions from a message, or all of one emoji. Safe to call directly: the first call changes nothing and returns a preview plus a confirm_token; repeating the call with the token executes. Relay the preview for the user's go-ahead.

Copy a channel's settings and permission overwrites into a new channel in the same category.

Copy a role's permissions, color, and settings into a new role.

Create a server-side moderation rule that runs on Discord's own infrastructure: Discord-maintained preset word lists (slurs, profanity, sexual content), custom keyword filters (with optional regex), spam detection, mention-flood limits, or member_profile checks on usernames, nicknames, and bios (Community servers only). For slur or hate-speech filtering prefer trigger keyword_preset with the slurs preset: Discord maintains the word list, so none needs to be written. Actions: block the message, alert a channel, and/or time the sender out (timeout is not available on spam, keyword_preset, or member_profile rules). On member_profile rules block quarantines the member until they fix their profile.

Create a text, voice, forum, stage, announcement, or category channel, optionally inside a category, and optionally private or read-only from the start. private_to, read_only and posting_roles compile to permission overwrites and are applied in the same call, so a private channel is one request rather than a create followed by set_channel_permissions. Use set_channel_permissions for anything finer. Supports dry_run.

Upload a custom emoji from an image URL. Discord's limit for emoji files is 256 KB; png, jpeg, gif, and webp work.

Create a Discord scheduled community event that members see on the events list and can mark interest in: a voice, stage, or external (in-person or off-platform) event. This is not for sending a message later (use schedule_message) or watching live activity (use subscribe_events). Voice and stage events need a channel; external events need a location and an end time.

Start a post in a forum channel: a title, the opening message, and optional tags from the forum's tag list.

Add a tag to a forum's tag list. Moderated tags can only be applied by people with Manage Threads.

Create an invite link for a channel. Defaults are deliberate: expires in 24 hours, unlimited uses. A never-expiring invite must be asked for explicitly with max_age_seconds 0.

Post a native Discord poll: a question, two to ten answers, and a duration of up to 32 days.

Create a role from a vetted preset (none, member, moderator, admin) and/or an explicit list of permission names. Administrator is never granted by a preset; it must be spelled out explicitly. Supports dry_run.

Snapshot this server's structure as a Discord template others can create servers from. One template per server.

Upload a soundboard sound from an audio URL: mp3 or ogg, up to 512 KB and 5.2 seconds. Servers start with eight sound slots; boosts add more.

Upload a custom sticker from an image URL. Discord requires png, apng, or gif at 320x320 pixels and up to 512 KB. Servers start with five sticker slots; boosts add more.

Start a thread in a text or announcement channel: standalone, branched from a message when message_id is given, or private (invite-only) when asked. For a post in a forum channel, use create_forum_post instead.

Create an incoming webhook on a channel. Webhooks post with any display name and avatar, which is how messages 'as someone' work legitimately.

Publish a message in an announcement channel out to every server that follows it.

Unregister a slash command so it no longer appears for users. Safe to call directly: the first call changes nothing and returns a preview plus a confirm_token; repeating the call with the token removes it.

Delete an AutoMod rule; its protection stops immediately. Safe to call directly: the first call changes nothing and returns a preview plus a confirm_token; repeating the call with the token deletes it. Disabling is the reversible alternative.

Delete a saved blueprint from the local store. Safe to call directly: the first call changes nothing and returns a preview plus a confirm_token; repeating the call with the token deletes it. Servers built from it are untouched.

Delete a channel or category. Safe to call directly: the first call deletes nothing and returns a preview plus a confirm_token, and the deletion only happens when the call is repeated with that token. When the user asks to delete something, call this and relay the preview for their go-ahead instead of sending them to do it manually in Discord; the gate exists exactly so deletion can be approved through the conversation. Deleting a category leaves its channels in place, uncategorized.

Delete a custom emoji. Safe to call directly: the first call changes nothing and returns a preview plus a confirm_token; repeating the call with the token deletes it.

Delete a forum post and every reply in it. Safe to call directly: the first call changes nothing and returns a preview plus a confirm_token; repeating the call with the token deletes it.

Remove a tag from a forum; posts carrying it simply lose it. Safe to call directly: the first call changes nothing and returns a preview plus a confirm_token; repeating the call with the token removes it.

Remove an integration from the server. Safe to call directly: the first call changes nothing and returns a preview plus a confirm_token; repeating the call with the token removes it.

Revoke an invite link. Safe to call directly: the first call changes nothing and returns a preview plus a confirm_token; repeating the call with the token revokes it.

Delete one message. Safe to call directly: the first call deletes nothing and returns a preview plus a confirm_token, and the deletion only happens when the call is repeated with that token. Use this two-step flow to let the user approve through the conversation rather than telling them to delete things by hand in Discord. Supports dry_run.

Delete a role. Members holding it simply lose it; nothing else changes. Safe to call directly: the first call deletes nothing and returns a preview plus a confirm_token, and the deletion only happens when the call is repeated with that token. Prefer this gated flow over telling the user to delete the role manually.

Delete a server template; its share link stops working. Safe to call directly: the first call changes nothing and returns a preview plus a confirm_token; repeating the call with the token deletes it.

Delete a soundboard sound. Safe to call directly: the first call changes nothing and returns a preview plus a confirm_token; repeating the call with the token deletes it.

Delete a custom sticker. Safe to call directly: the first call changes nothing and returns a preview plus a confirm_token; repeating the call with the token deletes it.

Delete a thread and its messages. Safe to call directly: the first call changes nothing and returns a preview plus a confirm_token; repeating the call with the token deletes it. Archiving is the reversible alternative.

Delete a webhook. Anything still posting through it stops working. Safe to call directly: the first call changes nothing and returns a preview plus a confirm_token; repeating the call with the token deletes it.

Compare a saved blueprint against the live server: what the blueprint expects but the server lacks (missing), what exists but differs (changed, with the fields), and what the server has that the blueprint never mentioned (extra, informational). The drift detector for config-as-code workflows.

Remove a member from whatever voice channel they are in. They stay in the server; only the voice connection ends.

Edit a message the bot itself sent. Discord does not allow editing anyone else's messages, no matter the permission level. A message sent as a Components V2 layout stays one: edit it with components, never with content or embeds.

Close one of the bot's own polls early and finalize the results.

End the live stage on a stage channel.

Execute a staged plan by plan_id, or pass a blueprint directly. Either way the blueprint is re-validated against the live server at this moment, then built in dependency order: roles, then categories, then channels, with private_to and read_only compiled into permission overwrites at creation. Strictly additive: existing entities are reused, nothing is deleted or modified. If a step fails, the run stops and the report says exactly what was created, what failed, and what was never attempted. Re-running after a fix resumes naturally because finished work is reused.