Server configuration: the channels the bot is pointed at, your server
rules, and the moderation policy you configure (rules, action preferences, retention,
excluded channels).
Message content: text of messages in channels the bot scans is cached
temporarily so rules can be evaluated against it (on servers that turn on ticket triage, up to 14 days, never past the retention setting, so a ticket can be checked against it). Message text that a decision flagged is
kept as a snapshot in the moderation record. The hour of chat after a public moderation
comment is sampled the same way: kept excerpts stay with that case.
Member data: Discord user and member IDs, and a running per-member
"vibe" score and rule-violation counts derived from sampled messages.
Moderation record: cases, the action taken (timeout, kick, ban, or
warning), who approved it, and the decision text shown to the member.
Classifier scores: for messages the first screening pass scores as
possibly breaking a rule, the text and the classifier's scores, at most 300 per rule per
server. Admins use them to preview how threshold changes would behave.
Moderator decisions: when a moderator approves, overrides or dismisses
a decision, a record of what was proposed, what the moderator chose, and the evidence behind
it (the messages involved, the rule text, and the classifier scores).
Billing: for paid plans, the server ID, the email address you paid with, the plan and its paid-through
date, and Stripe's customer and subscription IDs. Card details are entered on Stripe and
never reach us.
Tickets and private threads: when a member opens a ticket, who opened
it, when, the private thread it opened and the optional reason they typed; for each member
Jani has written to privately, the ID of that member's private thread. The threads themselves
live in Discord, not with us. When a server turns on ticket triage, Jani reads the ticket and
the server messages it looks up (sent to the AI model like any decision) and keeps the
ticket's kind, urgency and a short summary for the moderators.
Dashboard sessions: single-use login links from
/dashboard. Only a hash of the link token is stored; no passwords exist.
Why we collect it
Solely to provide the moderation service: screening messages against your server's rules,
deciding consequences, reporting to your moderators, and maintaining the audit trail a
moderation bot is expected to keep. Moderator decision records are also used to measure and
improve the service's moderation accuracy, including training or tuning the models it
uses. The bot does not sell data and does not use it for
advertising or profiling outside the service.
How long data is kept
Cached message text is purged daily once it is older than the server's
retention setting — default 30 days, settable 1–365 days per server in the
dashboard.
Ticket reasons and triage summaries are blanked on the same retention setting; the record that a
ticket was opened stays with the moderation record.
Classifier scores keep at most 300 messages per rule per server, oldest
replaced first; their message text is purged on the same retention setting.
Moderator decision records are kept indefinitely, including after the bot
is removed, because they are the data used to improve moderation accuracy. To have a
server's records deleted, contact us (below).
Billing records are kept as long as tax and accounting law requires.
Violation snapshots, cases, decisions, vibe scores, and after-action excerpts
are retained as the audit trail for the moderation record, until the server changes its
configuration or removes the bot. After-action excerpts are not blanked when cached message
text is purged.
After a bot is removed from a server, processing of that server stops immediately.
Third parties
Discord — the platform itself. Messages and member data originate from
and are delivered back to Discord. Discord's own terms and privacy policy also apply in
Discord servers.
OpenRouter — message content (flagged messages and their surrounding
context) is sent to AI models routed through OpenRouter for classification and moderation
decisions.
Stripe — processes payments for paid plans. Stripe receives the
payment details you enter at checkout under its own privacy policy.
Supabase — the bot's database (configurations, cached messages,
moderation records) is hosted on Supabase.
Control, removal and questions
To stop processing: run /pause (or use the pause switch in the dashboard), or
remove the bot from the server — removal stops all processing immediately. How long cached
message text is kept is under your control in the dashboard's retention setting. To delete
configuration and stored records for a removed server, or if you have a privacy question,
email support@jevetica.com with the server ID.