July 16, 2026
Five features now on every server
- NewNew-joiner probation has finished rolling out — when a brand-new member’s very first posts are flagged, they’re automatically paused from posting until a moderator takes a look. Raid-resistant by default, human-decided as always.
- NewSticker screening has finished rolling out — sticker images go through the same review as posted images, and a clean sticker post is left completely untouched.
- NewAnonymous member reporting has finished rolling out — right-click a message and Report to mods, and the report reaches your mod channel without exposing who filed it.
- NewThe keyword blocklist has finished rolling out — terms you configure are caught instantly, before any AI runs, and still go to your mods for the final call.
- NewThe graduated penalty ladder has finished rolling out — configurable warn and timeout rungs before anyone is banned, with strikes that can decay as members keep a clean record.
- ImprovedWhen a moderator overturns an appeal, the bot now also withdraws what it learned from the original rejection — so a message a human just cleared can't be auto-rejected again as a "repeat offense" if it's posted again.
July 15, 2026
MessModAI is now Docent
- NewNew name, same bot: MessModAI is now Docent — a calm guide that keeps watch so your moderators don’t have to. Nothing changes in your server: same settings, same permissions, same human-in-charge moderation. You’ll see the new name and logo in the dashboard, in notices, and here on the site.
July 13, 2026
Early-access capacity limit
- NewDuring early access, a member-count limit may apply to newly joining servers so moderation quality stays high while we scale. Over-limit servers get a polite notice and the bot leaves — nothing is set up and no data is kept. Larger communities can get in touch to be pre-approved.
July 12, 2026
Mod-log channel feed & server analytics
- Rolling outCustom notice copy templates (Settings → Notice Copy Templates): rewrite the body of the DMs members receive — under review, removed, warned, paused, banned — in your own words or your own language, with variables like
{{guild_name}} and {{rules}} filled in automatically. Leave a field blank to keep the standard message. Reaching servers gradually.
- NewAn Analytics tab on your server dashboard: messages scanned, flags per day, top flag categories, your moderators’ overturn rate and response time, and month-to-date AI usage — over the window your plan’s retention keeps.
- NewAn optional mod-log channel (Settings → Mod Log Channel): every resolution — approvals, rejections, warnings, withdrawals, appeal outcomes, pauses, timeouts, and bans — posts one compact line to a channel you pick, so your team has a single scrollable audit feed. Off unless you choose a channel.
- NewAn optional weekly moderation digest: once a week the bot posts a "your week in moderation" summary to your mod channel — messages scanned, flags, decisions, overturn rate, response time, top categories. The same numbers as the Analytics tab, no AI involved. Off by default.
- ImprovedThe dashboard's Observation Log, Flagged Messages, and User Strikes tabs are now one Activity tab: every scan verdict and every case in a single timeline with one filter row (All · Flagged · Cleared · Under review · Approved · Rejected), one detail view, and strikes right below. Old links redirect to the matching filter.
- ImprovedCommunity Flags and AI Context merged into one AI Memory tab — everything the AI learns from in a single list with a source filter, and the read-only feedback summary now sits next to the live digest with their difference spelled out. Old links redirect.
- Rolling outOne ❌ Reject button on mod review cards instead of two: it opens the rules picker with the AI's citation already selected, so agreeing with the AI is Reject → Confirm — two clicks, zero reading. Picking different rules works exactly as before. Reaching servers gradually.
- ImprovedMod review cards now speak moderator language — "safety filter (high certainty)", "AI review", "matched a prior moderator decision" — instead of internal layer numbers and vendor names. Technical provenance stays available in the dashboard's Activity details.
- NewThe moderator reject-picker can now use your Community Rules directly (Settings → Rejection Rules) — no more maintaining the same list twice. On by default for new servers; existing servers keep their custom list until they opt in.
- ImprovedSettings got simpler: the three auto-pause protections are one card with shared controls, appeals are a single Off / External link / In-Discord choice, reporter reliability tucks inside Community Flagging, and the two unrelated "Warn" features are now named apart. Your saved settings are unchanged.
- ImprovedSettings reorganized into five sections — Core, Rules & AI, Enforcement, Spam & Flood Protection, Community & Workflow — with long explanations tucked behind "ⓘ Learn more" disclosures. Every control keeps a one-line description, and consequence notes (strikes, bypasses) stay visible. Nothing about your saved settings changed.
July 11, 2026
Link policy: domain blocklist & invite-link control
- Rolling outA per-server link domain blocklist — messages linking to a domain you list are removed instantly, before any AI runs, subdomains and paths included, and still go to your mods for the final call. Paste full URLs into Settings if that’s easier; only the domain is kept.
- Rolling outAn optional block Discord invite links switch — invite links are removed and sent to mod review. Exempt your welcome channels if your own team posts invites there. Off unless you turn it on.
July 10, 2026
New-joiner probation, docs & pricing pages
- Rolling outNew-joiner probation — when a brand-new member’s very first posts are flagged, they’re automatically paused from posting until a moderator takes a look. Raid-resistant by default, human-decided as always. Rolling out gradually to servers.
- Rolling outSticker screening — sticker images now go through the same review as posted images, so a rule-breaking sticker can’t slip past just because it isn’t an upload. A clean sticker post is left completely untouched. Reaching servers gradually.
- NewA documentation page: the three-step setup guide and a moderator handbook covering mod cards, cool-off holds, strikes, and appeals.
- NewA pricing page — the planned tier structure, and everything free during early access.
- NewThis changelog, so you can always see what the version moderating your server actually shipped.
July 9, 2026
Appeals, slash commands & member reporting
- NewAppeals (opt-in in Settings): rejection and ban notices carry an ⚖️ Appeal button. The member writes their reasoning, a fresh card reaches your mod channel, and overturning reverses the strike — or lifts the ban. Appeals nobody reviews within 14 days (sooner on servers with short log-retention settings) close as upheld, and the member is told.
- NewSlash commands for moderators: /history and /strikes to look members up, /pause and /unpause for posting pauses, /exempt-channel and /observation for quick settings — plus a right-click View User History shortcut.
- Rolling outAnonymous member reporting — right-click a message and Report to mods, and the report reaches your mod channel without exposing who filed it. Reaching servers gradually.
- NewThe MessModAI website you’re reading launched, on the same domain as the dashboard.
- Rolling outA per-server keyword blocklist — terms you configure are caught instantly, before any AI runs, and still go to your mods for the final call.
- Rolling outA graduated penalty ladder — configurable warn and timeout rungs before anyone is banned, with strikes that can decay as members keep a clean record.
- FixedImage posts in servers with image review no longer briefly disappear while a clean image is being checked (rolling out gradually), and image analysis is more reliable.
Earlier improvements predate this changelog. Questions about a release? [email protected]