Privacy Policy

1. What we collect

The Service processes messages in memory only in order to scan for policy violations. The following data is written to the bot's database (PostgreSQL, run by our hosting provider) when a relevant event occurs:

• Violations: when a message triggers the filter: Discord user ID, username at the time, channel ID, server ID, the matched term, a copy of the message content, a confidence score, and a timestamp.
• Warnings: when a moderator issues a warning: the warned user's ID, the moderator's ID and name, reason text, channel ID, and a timestamp.
• False-positive reports: when a user or moderator flags an incorrect match: user ID, username, the flagged word, the original phrase, and a timestamp.
• Voice activity: when a member joins or leaves a voice channel: user ID, channel ID, server ID, join time, and session duration (in seconds). No audio is retained.
• Bot usage summaries - when a user actually uses the bot or is scanned by it: Discord user ID, username at the time, server ID, channel ID, first/last seen timestamps, last command/component name, and aggregate usage counters. Message text is not stored in these summaries.
• Admin & whitelist IDs: Discord user/role/channel IDs that have been granted admin access or added to the whitelist via bot commands.
• Learned patterns: normalised word roots (no usernames or IDs) derived from violation text by the self-learning engine.
• Chat generator statistics: only in channels where a server admin has turned on /chatter: counts of which word followed which pair of words in that channel, used to post new messages in the same style. The messages themselves and who sent them are not stored, and the counts cannot be traced back to a person. Off by default.
• Unique-message fingerprints: only in channels where a server admin has turned on unique messages: a one-way hash of each message said there, used to spot a repeat. The text and who sent it are not stored. Off by default.
• Writing-style fingerprints: only in servers where an admin has turned on behavioral fingerprinting: for each member, their Discord ID, message count, average message length, punctuation rate and 20 most-used words. When a member is banned, theirs is kept so a new account writing the same way can be flagged to staff. Off by default.
• Moderation cases: timeouts, kicks, bans, unbans, and moderator notes: the affected user's ID and username, the moderator's ID and name, reason text, and a timestamp.
• Server change records: in servers with premium: each channel, role, permission and ban change, the Discord ID and name of who made it, and the IDs of members who held a deleted role. Kept for the number of days the server owner sets, 30 by default and 90 at most.
• Tickets & appeals: support ticket subjects, status, and who claimed/closed them, plus a transcript saved when a ticket is closed: each message in the ticket channel with its author's name, ID, text and time, so staff can review the ticket after the channel is deleted; ban appeal text and its review outcome. Tied to the submitting user's ID.
• No-contact orders: the Discord IDs of the two users placed under a no-contact order and any nicknames/aliases a moderator adds for name-mention detection.
• Premium status: for servers with premium: the tier, who granted it, and its expiry date. No payment details are stored (see section 5).
• Heat scores: when a member triggers the filter: their Discord ID, the server ID, a number that rises with each violation and falls over time, and the last automatic action taken. Used to escalate repeat offenders.
• Cross-server ban records: when a member is banned from a server running the bot: their Discord ID, that server's ID and the time. How many servers have banned someone is shared with every server running the bot: it can warn staff when that person joins, and it makes the filter escalate faster for them.
• Saved roles: only in servers where an admin has turned on role persistence: the role IDs a member had when they left, so they can be given back if they rejoin.
• Timeout records: when a member leaves while timed out: their Discord ID, the server ID and when the timeout ends, so it is applied again if they rejoin before then.
• Message activity: for server activity statistics: the Discord ID, channel ID and time of each message. The message text is not stored.
• Modmail and staff applications: the user's ID and name, the staff thread, its status, and who closed it. The conversation itself stays in Discord.
• Bug reports and suggestions: the text sent with /bugreport or /suggest, with the sender's ID and name and the server it came from.
• Whips: the member who set one up, its target, and the trigger and reply text.
• Whodunnit guesses: when a member sends a guess in a server's Whodunnit game: their Discord ID and name and what they typed. Only the people the game's host picks see it: a staff channel, the host, and any members or role the host chooses to DM each guess to.

Outside ticket transcripts and text you give the bot on purpose (appeals, bug reports, suggestions, whips), message content that does not trigger a match is never written to disk.

2. Why we collect it

• Violation records let moderators review what was caught and appeal decisions.
• Warning history lets moderators track repeat offenders.
• Voice activity is used only for the /activity server-statistics command.
• Bot usage summaries show who actually used or was processed by the bot without storing message content.
• Learned patterns improve detection accuracy over time.

We do not use any data for advertising, analytics, or sale.

For users in the European Economic Area or United Kingdom: we process this data on the basis of legitimate interest: specifically, the legitimate interest of you and your server's admins in maintaining a safe community and enforcing your server's rules.

3. How long data is retained

We automatically delete the following on a rolling basis, regardless of whether the bot is still in your server:

• Violation records: after 90 days.
• Voice/server activity records: after 90 days.
• Ban appeals: after 180 days.
• Moderation cases (timeouts, kicks, bans, notes): after 365 days.
• Chat generator statistics: kept while /chatter is on in the channel, deleted immediately by /chatter disable or /chatter forget.
• Unique-message fingerprints: kept while unique messages is on in the channel, deleted immediately when it is turned off.
• Writing-style fingerprints: dropped after 7 days without a message from that member. A banned member's is kept while the bot remains in that server.
• Heat scores: dropped once they have cooled off.
• Cross-server ban records: removed for a server when it unbans the person.
• Saved roles and timeout records: deleted when the member rejoins and they are applied.
• Message activity: after 90 days.
• Whodunnit guesses: deleted with their game. Staff can delete a game at any time, and a server keeps its last 5 games: past that, the oldest finished game and its guesses are deleted.

Warnings, tickets, no-contact orders, admin/whitelist IDs, and premium status are not subject to a fixed retention period and persist as long as the bot remains in your server, since they're actively relied on for ongoing moderation. Server owners and admins can delete individual warnings, whitelists, and other records directly using bot commands or the dashboard where those controls are available. Data you'd like deleted sooner, or data associated with a server that has removed the bot, can be requested for deletion by contacting us (see section 11).

4. Who can access your data

• Server admins & owners: can view and delete violations, warnings, and whitelists for their server via bot commands and the web dashboard.
• Bot owner: has access to all stored data for maintenance and debugging purposes only.
• Third parties: data is never shared, sold, or disclosed except where required by law.

5. Third-party services

• Discord: the bot operates on Discord's platform. Discord's Privacy Policy governs data they hold.
• Google Translate: when translation scanning is enabled, non-English message text is sent to Google Translate. Message text sent to this API is subject to Google's Privacy Policy. Server owners can turn translation scanning off for their server in the dashboard's server settings.
• Google Gemini: when AI context checking is enabled, a message that the filter has already flagged is sent to Google's Gemini API, together with the term that matched, so the model can confirm or overturn the detection. Because only flagged messages are sent, that text can contain slurs or other sensitive wording. Message text sent to this API is subject to Google's Privacy Policy. Off by default and controlled by the bot operator.
• Sightengine: when a server admin turns on AI image detection and the bot operator has set up Sightengine, images posted in that server are sent to Sightengine to check whether they are AI-generated. Off by default. NSFW image scanning is separate and runs on the bot's own server: those images are not sent anywhere.
• Ko-fi / Stripe: premium purchases are processed entirely on Ko-fi's or Stripe's own checkout pages. We never receive or store payment card details, only a webhook notification confirming the guild ID, tier, and duration purchased.
• Anthropic: if the operator has enabled automatic crash reporting, an unhandled error sends the stack trace and recent host logs to a Claude Code routine so the fault can be investigated. These reports are about the software, not about you, but a trace can incidentally include fragments of the message being processed when the error occurred. Off unless the operator configures it.
• GitHub: with automatic crash reporting on, the same crash and error reports are also filed as issues in the bot's private GitHub repository, which only the operator can read. Like the reports above, a log excerpt can incidentally include a fragment of a message. Off unless the operator configures it.
• Backup storage: if the operator sets up off-site backups, a nightly copy of the bot's database is uploaded to an S3-compatible storage provider of the operator's choosing, so the data survives the loss of the main server. Off unless the operator configures it.

6. International data transfers

Your data may be processed or stored in a country other than your own, including the United States, where the bot's hosting provider and the third-party services listed in section 5 (Google Translate, Google Gemini, Sightengine) operate. Where required, we rely on the transfer safeguards those providers make available (such as their own compliance with recognised data-protection frameworks) to protect data moved across borders.

7. Security

The database is run by the bot's hosting provider. The web dashboard requires Discord OAuth authentication and uses signed, HTTP-only session cookies. We take reasonable steps to protect stored data but cannot guarantee absolute security.

8. Cookies

The dashboard sets a single session cookie when you log in with Discord OAuth. It's strictly functional: it identifies your logged-in session so the dashboard knows which server(s) you can manage, and is not used for tracking, analytics, or advertising. It's cleared when you log out or expires automatically after a period of inactivity. We do not use any other cookies.

9. Data breach notification

If we become aware of a security breach that compromises your personal data, we will notify affected server owners and/or users without undue delay, as required by applicable law.

10. Children's privacy

The Service is not directed at children under 13 (see the age requirement in our Terms of Service), and we do not knowingly collect data from anyone under that age. If you believe a child under 13 has provided data through the Service, contact us (see section 11) so we can delete it.

11. Your rights & contact

For the purposes of applicable data-protection law, the bot owner is the data controller for the data described in this policy.

You may request the following by contacting us: deletion of your personal data (violation records, warnings, voice activity, and bot usage summaries associated with your user ID); access to a copy of it; correction of inaccurate data; and, where applicable under EU/UK GDPR, restriction of, or objection to, our processing of it, and portability (a machine-readable copy to transfer elsewhere). Server owners may also use bot commands to manage records within their own server without contacting us.

California residents: we do not sell or share your personal information, so there is nothing to opt out of. You may still request access to or deletion of your data as described above.

Contact: legendzz749@gmail.com

12. Changes to this policy

We may update this policy. Material changes will be announced in the support server. Continued use of the bot after changes constitutes acceptance.