Server AssistantServer Assistant Features
Each feature area pairs what you get free (the green cards) with what Trial & Premium add (the gold cards). Tap any card to expand the detail. See also Commands and Pricing.
⛏️ Minecraft + Discord, one community Free on every plan
MCDC bridges your Minecraft server (Paper/Spigot 1.21.x, Java 21) and your Discord into one community, with one moderation standard across both. Set it up with the single /mcdc wizard in Discord, then link the plugin with one /mcdc link command at your server console (or paste the same details into config.yml), no restart needed.
- Two-way chat: players post as themselves, real name and skin avatar, and Discord reactions show up in game.
- Server events in Discord: join, leave, death, advancement, start and stop, each one toggleable.
- Live health: TPS, RAM and uptime in Discord, a live player count and health dot in the channel topic,
/online, plus optional offline and low-TPS alerts. - Playtime XP into the same shared balance as Discord activity (AFK time excluded).
- Moderation on every surface players write on: chat, signs, books, name tags, anvil renames,
/me, private messages, and usernames on join, via the same AutoMod switch as Discord./mcbanand/mckickwork by player name. - Web ban appeals, even for bans Server Assistant didn't issue: console or other-plugin bans are captured and get an appeal link, shown on the ban screen and again if the banned player tries to rejoin.
- One-click account linking between a Discord member and their Minecraft player.
- Discord roles mirror to Minecraft ranks (one-way, via LuckPerms), and the plugin self-updates with sha256 verification.
Premium extras @sai in game, grounded in the real Minecraft wiki with clickable citations · a personal Allay companion that answers privately · a per-server persona (name, tone, attitude) · live two-way chat translation.
Locked down by design: outbound HTTPS only, no inbound ports, no RCON, and the token is scoped to exactly one guild and channel.
Unlock the full toolkit
Get these three popular Premium features, 750K AI tokens a month, plus everything else listed below.
SAi assistant
Your AI co-pilot, it reads your server's setup and tunes it for you, right from the web portal.
White-label branding
Completely rebrand the bot as your own, your name, icon and colour on every member-facing embed.
AutoMod AI second-opinion
Borderline filter hits get an AI confidence check first, so fewer good messages get caught and staff review only the close calls.
Ready to upgrade? Run
/premium in your server.
Trial features use your AI tokens (free trial included); Premium needs a subscription.
Free
Trial & Premium
Moderation
Role-based tiers (customisable)
Defaults: Owner / Admin / Moderator. Each tier has its own capabilities + approval authority, fully customisable in
/settings → Role Tiers (add Helper, Trial Mod, Senior Mod, anything).Tiered approval (actions route up)
Dangerous actions (like
/purge) route up a tier: admins approve moderators; the owner approves admins. Anyone with approval authority, or the owner, runs them directly.Persistent warnings + DMs (/warn)
Every warning is timestamped, attributed, and the user gets a DM with their current count and escalation context. Run
/warnings @user or /manage-warnings to view or remove.Private staff notes (/note)
For tracking behaviour without a formal warning. Invisible to the user; staff-only via
/note, /notes, /manage-notes.Soft-ban + timed purge (/softban · /purge)
/softban ban-then-unban to sweep messages without permanent removal. /purge 50 or text purge since 1pm (your timezone; capped at Discord's 14-day limit).Smart purge (/purge mode: Smart)
/purge … mode: Smart, AI cleanup that removes only the genuinely problematic messages and keeps the rest, instead of clearing everything like a plain purge.Timed bans (/tempban)
/tempban @user 7d reason bans for a set time, 30m, 2h, 7d, 1w, even 1d12h (1 min to 28 days), then un-bans automatically. The timer survives restarts, and the lift is logged.One-command roles (/role)
/role add @user @role / /role remove @user @role, no digging through Discord's menus. Guards against privilege escalation: you can only assign roles below your own, and managed/integration roles are off-limits.Server-wide lockdown (/lockdown)
/lockdown start locks every text and forum channel in one shot, including their threads and posts. /lockdown end restores each channel to its exact prior permissions, not a best guess.Repeat-offender detection (/altguard)
Enable with
/altguard on. Fingerprints the users you ban or kick, then checks every new joiner, a fresh account reusing a banned user's avatar or near-identical name is auto-banned and reported to staff; weaker matches are flagged. Soft-bans excluded.24-hour undo
Reverses your most recent ban, mute, channel deletion (incl. permissions), or nickname change. Note: deleted message contents can't be recovered, that's a Discord limit.
Bulk via role mention
mute @TrollRole 1 hour actions every member of that role at once. Bots auto-excluded; >5 targets requires the Owner.Ban reasons & appeals
For the banned member (reason + one reply)
Every ban (manual,
/tempban, AutoMod, or native) DMs the user the exact reason staff supplied, plus an invite to send one reply to appeal. Just one, no escalating threads, no DM noise. Soft-bans excluded, and so are ThreatNet auto-bans, which are silent and open no appeal ticket (that automated decision is contested via the web portal instead).For your staff (one-tap appeal ticket)
That single reply opens an appeal ticket in your staff channel with one-tap buttons:
| Button | What it does |
|---|---|
| Unban | Lifts the ban, DMs the user, writes to the audit trail. |
| Deny | DMs the denial, closes the ticket. |
| Research | Runs Message Report on their last message for context. |
| Info | Profile, dates, roles, warning + note counts, Threat Score. |
| Warnings | Full history, even if they've left. |
| Notes | Staff notes (private). |
No AI token is spent unless Research is pressed. On by default; buttons survive restarts.
"Why was I actioned?"
When a member is warned or muted, the bot DMs a short notice with a Why was I actioned? button that returns a calm, plain-language explanation from the staff-supplied reason. They can also DM the bot "why was I actioned?" for the same on their most recent warning.
Guided multi-round appeals
A ban appeal becomes a short AI-facilitated back-and-forth: the member shares their side, the bot asks one or two neutral follow-ups, then hands staff a concise summary plus the full transcript, on the ticket and the portal. You set the max rounds (default 3); the facilitator never decides or promises an outcome.
AutoMod Engine
Filter presets (Off → Strict)
One-click packs in /settings → Quick Presets → AutoMod:
| Preset | Catches |
|---|---|
| Off | Nothing |
| Scams only | Free Nitro / Steam gift / Robux / IP-grabber links |
| Scams + slurs (recommended) | Above + Level-1 hate speech. Allows casual profanity |
| Strict | Above + slurs L2, profanity L3 (child-friendly only) |
Server-type templates (/setup)
The final /setup step tunes moderation to the kind of server you run, a combo of AutoMod pack + punishment ladder + anti-raid sensitivity.
| Type | Tuning |
|---|---|
| General | Balanced defaults for a mixed community. |
| Gaming | Scams + slurs, allows casual profanity. |
| Creator | Stricter anti-raid for follow-raids. |
| Anime / Fandom | Community-friendly, high-energy chat. |
| Crypto / Trading | Strict anti-scam, fast escalation. |
| Education | Gentle, warning-first, coaching. |
| NSFW / 18+ | Stricter anti-raid for adult communities. |
| Manual | Applies nothing, full control. |
Pre-selected from your server's name. Applied only on a fresh setup or deliberate pick, so re-running /setup never overwrites your tuning.
Custom regex filters (/automod)
/automod → Add Filter, with a built-in safety check for catastrophic-backtracking patterns before a rule goes live.Link blocklist (/automod)
Domain-level blocks, with known IP-grabber domains preloaded.
Punishment ladder
Auto-escalation: Gentle 5/10/20 · Standard 3/5/10 · Strict 2/3/5 (warns → timeout → kick → ban).
Anti-spam
Auto-on; 5+ duplicate messages in 5 seconds are removed and the user is warned.
User DM on removal
Every AutoMod removal DMs the user which rule fired plus their warning count. Staff see the full embed in the log channel either way.
Self-trained AutoMod (learns from you) Free trial
Watches the messages your staff delete or report, then proposes server-specific AutoMod rules for your approval. Nothing auto-enforced; 30-day local log, no identities stored. Enable in
/automod → Self-training.AutoMod AI second-opinion
Optional in
/automod → AI Review. Word-filter and spam hits get an AI confidence check first: confident ones act, low-confidence ones go to a one-tap staff review card (Delete & warn / Allow) instead of an auto-delete. Link and mention hits are not sent. Cuts false positives; word-filter checks run on a free moderation endpoint and cost you nothing, spam checks use your AI allowance.Server health & insight
Pulse (whole-server health digest) Free trial
A one-embed health digest, activity counts, a mod-team recap, a reception summary, with week-over-week deltas and flags for the quiet patterns humans miss. Always on in the web portal for every server; each staff member can also opt in to a personal DM, Off / Weekly / Monthly, under Settings → Personalization in the portal.
Milestones (celebrate your wins)
Server Assistant celebrates your community's wins in staff chat with a warm card, member-count landmarks (100, 500, 1k, 5k, 10k…), server anniversaries, and moderation milestones, framed as your achievement. Capped to at most one a day, and fully toggleable in
/settings → Notifications.Bot Health Insurance (it watches itself) Free trial
The bot watches its own footprint. If it gets heavy-handed, it suggests tuning, at most once every 14 days, so the anti-noise feature never becomes noise. Toggle in
/settings → Behavior.Threat Score (per-user risk read, this server)
At-a-glance per-user risk, Low / Elevated / High, in user profiles, weighted by severity and decaying over time. Advisory only: it informs staff and never takes an action by itself. Full Premium (not BYOK), and off until you switch it on in
/settings → Behavior. (For cross-server history, see ThreatNet below, whose advisory band is free for every server.)Cross-server safety (ThreatNet)
ThreatNet (cross-server threat network)
Every protected server shares minimized, severity-only abuse signals, so a user with a serious, corroborated history of scams, raids or ban-evasion on other servers lights up on yours, ideally before they cause harm. Shared as aggregates and a generic severity level only, never the offence type, your staff's reasons, message content, or which server acted. A core, on-by-default safety feature for every server; individuals can opt out of profiling in the web portal, subject to a published safety exception. The advisory band is free for every server.
Scam-image defense (catches scams hiding in pictures)
Lots of scams are a screenshot, not text, a fake "withdrawal received" payout, a giveaway notice, a staged testimonial, so text filters miss them. ThreatNet now fingerprints posted images and removes ones matching a cross-server scam-image blocklist. On by default, every plan. A confirmed match removes the image and times the poster out; each server can soften this to quarantine or advisory (flag-only). On AI-enabled plans, images from brand-new members get an extra AI vision check for never-seen scams, which teach the network once corroborated. Only an opaque, one-way fingerprint of a flagged image is shared, never the picture, never linked to a user, under the same opt-out as the rest of ThreatNet.
ThreatNet, rich view (Premium)
Premium servers see the richer cross-server dossier, severity band, how many distinct servers corroborate it, and recency, surfaced alongside (never merged with) the local Threat Score on user profiles.
ThreatNet auto-protect (Premium · opt-in)
Close the gate before they act. Turn on auto-protect and a user who joins meeting a cross-server risk threshold you choose is banned automatically on join. The default is the highest band (serious and corroborated across 2+ independent servers, the safest); you can set a broader level, which acts on weaker signals and carries more risk of a mistake. Off by default, enabled per server with
/threatnet autoban on or in /settings → Security → ThreatNet (Manage Server). Because it's an automated action it ships with safeguards: opted-out users are never auto-banned, every auto-ban is in your audit log, and although the ban is silent, no DM and no in-server appeal ticket (unlike a normal ban), it's disclosed in our Privacy Policy and stays appealable via the web portal (human review). You choose the threshold and remain responsible for moderation in your server.AI tools
Image generation (/imagine) Free trial
Type
/imagine <prompt>, get a 1024×1024 image. Uses your own AI key (/ai-config); otherwise the free trial covers ~30 images and Premium is unlimited. 30s cooldown on the free tier.Message Report (context summaries) Free trial
Right-click any message → Apps ▸ Message Report for a 3 to 5 bullet summary: tone, possible rule-breaking, escalation, recommended action. Only you see it, with one-tap Warn · Timeout · Delete · Kick · Ban plus Info · Warnings · Notes (each respects your role). The bot never auto-acts, staff always decide.
Raid Protection
Anti-Raid Detection (join-rate alerts)
Tracks join rate. When the threshold trips, the bot pings the Owner (and any roles in /settings → Notifications), suspends onboarding DMs, and logs the event.
| Preset | Threshold |
|---|---|
| Off | None |
| Normal | 5 joins in 30s |
| Strict | 3 joins in 15s |
Reception verification (welcome desk on join)
Optional measure (
/settings → Verification). New members land at an in-server Reception desk with a Verify me button and a quick are-you-human check; passing grants the verified role. Posting instead of verifying trips its honeypot. Stops most automated raid bots cold.Onboarding & canned
Auto-onboarding DMs (/onboarding)
Welcome DMs (
/onboarding) with placeholders: [servername], [usermention], [membercount], [owner], [usercreatedat]. Templates: Friendly · Gaming · Professional · Minimal · Custom.Auto-role on join
Pick the role new members receive automatically (
/settings → New Members → Auto-role).Welcome reaction emoji
The bot reacts to "X joined the server" system messages with your chosen emoji.
Snippets (/snippets)
/snippets opens a panel to add and manage canned replies (supports the same placeholders as onboarding). Mention a user when invoking to fill [user] slots.Reminders (/schedule)
/schedule opens a modal to set a one-off or recurring reminder (e.g. "in 30 min" or "at 9am"). Uses your /timezone. Fires in the channel where it was created.Audit Logging
Colour-coded log (every action)
Every action posts a colour-coded embed to your log channel:
| Status | Colour |
|---|---|
| EXECUTING / SUCCESS / APPROVED | Green |
| OWNER_BYPASS | Blue |
| PENDING_APPROVAL | Yellow |
| DENIED / FORBIDDEN | Red |
| PERMISSION_DENIED | Orange |
Verbosity: All / Dangerous-only / Errors-only (/settings → Behavior).
Weekly digest
Every Monday at 9am (configurable day) the bot posts a summary: total actions, warnings, top action types, most active staff, member count. Separate from the Pulse health digest.
Tamper-proof trail
Native right-click bans/kicks/timeouts are logged like bot commands. Every action is written to a separate record encrypted at rest that staff can't read, edit, or erase; deleting a log-channel entry alerts the owner while the encrypted record survives. Opt into a locked
#server-assistant-log channel during setup.Permission health check
Once a day the bot verifies it still has Manage Roles plus a working staff + log channel. If an admin change broke any of them, the owner gets a one-time DM naming exactly what broke and how to fix it. Idempotent, quiet, read-only.
Customisation (/settings)
Branding
Embed colour (12 presets + custom hex), bot nickname, default timezone.
Quick Presets
One-click bundles for AutoMod, punishment ladder, onboarding template, anti-raid.
Behaviour
Approval timeout, log verbosity, weekly digest day, self-checks, Threat Score.
Role Tiers
Custom tiers and approval authority per tier.
Notifications hub (updates + instant alerts)
One place for every way the bot keeps you posted (/settings → Notifications), split in two:
| Section | Controls |
|---|---|
| Updates & digests | Toggle Milestones and Announcements on or off, your call. |
| Instant alerts | Which staff roles get pinged on raids / approval requests / AutoMod escalations, live, because you can't put a raid on a digest. |
Pulse isn't set here, it's always on in the web portal, where each staff member picks their own DM cadence (Off / Weekly / Monthly).
Channel allowlist
Choose where the bot responds to mentions.
Web Portal
Insights
Expandable KPIs (members · mod actions · AI usage) with trend charts, a member-growth chart, online / in-voice counts, a paginated activity log with real names, and your always-on Pulse health digest + weekly mod recap.
Act from the web
Warn / timeout / kick / ban / unban (per your role) and resolve staff decisions, ban appeals, AutoMod reviews, owner-approvals, raid lock-downs, from a "Needs input" strip, kept in sync with the Discord card.
Install + notify
Add the portal to your home screen as an app and turn on notifications to get a push the moment something needs your input.
SAi assistant
Your AI assistant in a chat bubble: ask in plain English and (as owner/admin) apply its suggestions in one tap.
Web setup
Brand-new servers can be configured entirely from the portal: pick your channels, roles and AI, and the bot does the rest.
Whole staff team
Owners, admins, and moderators can all sign in, and the portal mirrors each person's role exactly, never granting more than they have in Discord. Verified through Discord, re-checked on the server for every action.
Share Server Assistant (portal card or /share)
A "Share Server Assistant" card in your portal with your own personal invite link, hand it to a friend's server in one tap, and watch the running count of servers that joined through yours tick up. Your staff can also grab the install link straight from Discord with
/share (free tier).Milestone badges
The milestones your community has earned, member-count landmarks, anniversaries, moderation streaks, show as badges in your portal header, so the wins you've hit are always in view.
Ticket Panels
Ticket panels (/tickets)
An admin runs
/tickets setup (panel channel, category, support role, optional transcript channel). Members click Create Ticket to open a private ticket-0001-username channel visible only to them, the support role, and admins. On close, a .txt transcript is saved and the channel removed. Free tier: one intake question + editable panel.Ticket Panels, Premium extras (/tickets)
Up to 4 custom intake questions (
/tickets questions), custom welcome + close messages, and add users mid-ticket (/tickets add @user).Self-service & branding toolkit
Reaction roles (/rolepanel)
Self-service role buttons built from one wizard: pick roles from a dropdown, name the panel, post it. Members click to add/remove a role, and a My roles button opens a private menu with a ✓ next to roles they have. The bot only offers roles it can actually assign.
Custom slash commands (/customcmd)
Define your own real
/command (text or embed), e.g. /rules, /faq, from a wizard. Embed commands have an optional Live toggle: edit once and every copy already posted updates. Names validated; up to 25 per server.AI Q&A from your FAQ (/faq)
Members run
/faq with a question and the bot answers only from your curated entries, when the FAQ doesn't cover something it says so and points to staff, rather than inventing an answer. Counts toward AI usage, with a per-user cooldown.Translation (right-click or /translate)
Right-click any message → Translate, or use
/translate, into your Discord language or one you name. On-demand only, so there's no per-message cost or channel noise.Mediator (/mediate)
The wizard has the bot DM two members, collect each side privately, and post a neutral summary to staff, a chance to de-escalate before anyone reaches for a moderation action. It never takes sides or recommends punishment.
Server backup & restore (/backup)
Snapshot your roles, categories, channels and their exact permission overwrites. Restore is owner-only, confirmation-gated, and strictly additive, recreates anything missing and reapplies overwrites, never deletes.
White-label branding (/brand)
Completely rebrand Server Assistant as your own. Give it your own per-server nickname, your name + icon on every member-facing embed, your colour, and your name in the footers, so to your members it looks like a tool you built, not a third-party bot. Discord allows only one global bot identity, so this rebrands the nickname and output, not the underlying account.
Privacy & security
Privacy & security
AI keys and credentials are encrypted at rest, per-server data is isolated, and removing the bot wipes your encrypted secrets immediately. Retention rules are in the Privacy Policy.