Privacy Policy

When you sign in with Discord, we request only the identify and guilds permissions — your basic Discord identity and the list of servers you're in. We do not request your email address.

• Your Discord user ID is stored so we can associate your blueprints, settings, and actions with you.
• Your Discord access token is kept only in an encrypted, browser-only session cookie (it expires after 7 days) — it is never written to our database. We use it to ask Discord, in real time, which servers you can manage. Your server list is not stored in our database.
• Discord is both our sign-in provider and a data recipient: when you open your server picker, we send your token to Discord's API to fetch your manageable servers.

• Blueprints — the server structure you author (roles, channels, categories, permission settings, onboarding, automod rules, channel topics and pinned-post text, community settings, and references to icon/banner/emoji images by URL), plus Packet feature settings such as the role picker.
• Snapshots and apply history — automatic backups of your structure and a record of each apply (what changed, when, the mode, and the Discord ID of who triggered it).
• Activity log — a record of structure-changing actions (e.g. "applied," "created snapshot") with the Discord ID of who performed each.
• Advanced direct-call log — if you use the advanced "raw" Discord-API passthrough feature, we store the full request and response of those calls. Because these can contain arbitrary Discord data, we treat them as sensitive and keep them only briefly (see Retention).
• Feedback — if you submit feedback, we store your message, your Discord username and avatar at the time, the page you were on, and your browser/screen details, plus any replies — from our team, and any you send back, including a reply you send to our feedback message in a Discord DM.
• Preferences — your per-account settings (theme, default options).

Message content: When Packet manages a channel, it briefly reads recent and pinned messages only to recognize messages it previously posted itself (so it doesn't duplicate them). This is processed in memory and is not stored.

To assign roles automatically when someone joins (the Role Caller's auto-role feature), Packet uses Discord's Server Members access and is notified when a member joins a server Packet is in. It acts on that event in memory to grant the roles you configured — it does not store a member list or roster. Packet's role picker also lets members click to add or remove their own roles; those actions change roles on Discord and aren't kept as personal records.

Sticky roles (the optional "restore roles on rejoin" feature). Where a server enables sticky roles, Packet records which roles a member held when they leave, so it can put them back if the member rejoins. This is the one Role Caller feature that stores member-specific data, and only this: a member's user ID and the role IDs they had. It is deleted as soon as the member rejoins (their roles restored) and, if they never return, is pruned automatically after about 90 days. The feature is off by default.

Temporary roles (the optional /temprole feature). Where a server enables temporary roles, a moderator can grant a role that auto-expires. Packet stores the grant — the member's user ID, the role, who granted it, and when it expires — until the role is removed, then deletes the record. The feature is off by default.

Moderation cases (the optional Magistrate feature). Where a server enables moderation, a moderator can warn, mute, kick, or ban a member with the /mod commands. Packet stores a case record — the member's user ID, the moderator's ID, the action taken, the reason the moderator typed, and when — so moderators have history and the automatic escalation ladder works. When the server turns on the option to also track Discord's own moderation actions, Packet records the same kind of case for a ban / kick / timeout done through Discord's own menu, or a timeout applied by Discord's AutoMod, reading only the action, the target, and the reason from Discord's audit log. It also counts how often a member's messages trip AutoMod, and records a case only if that happens repeatedly in a short window — a count, never the messages themselves. It does not store any message content. The feature is off by default.

Activity status (the Activity roles feature). Where a server enables Packet's optional Activity roles feature, Packet uses Discord's Presence access. Through it, Discord shares a member's current activity — for example a game they're playing, or another app or service (such as music) shown in their Discord status. Packet compares that activity to the rules the server configured and adds or removes a role to match, in memory as the activity changes. It does not store any member's activity or status — the assigned Discord role is the only record. The feature is off by default and runs only where a server's admins enable it.

Support tickets (the optional Concierge feature). Where a server enables tickets, a member can open a private support channel from a panel. Packet stores ticket metadata — who opened it, the staff member who claimed or closed it, its status and timing, the close reason, and (if the panel asks for one) the short subject line the member types when opening. If a desk asks for feedback when a ticket closes, Packet stores the 1–5 star rating and the optional comment the member chooses to leave. To power support signals like "awaiting a reply" and first-response time, Packet sees that a message was sent in a ticket — who sent it and when — but not its text.

Separately, a server's admins can turn on an optional conversation view so support staff can read and reply to a ticket's thread from the dashboard — kept scoped to ticket channels only, for easy management of your support and your own record-keeping of those conversations. When a staff member opens a ticket, Packet reads that ticket channel's recent messages — including their text — live, on demand, only to display them, and stores none of it: an edit or deletion on Discord is reflected the next time the thread is loaded, and any attachments are shown as links only (Packet never downloads them). This uses Discord's "Message Content" permission, applies only to ticket channels, and is off unless a server's admins turn it on for their server. The tickets feature as a whole is off by default and runs only where a server's admins enable it.

Leveling and leaderboards (the optional Engagement feature). Where a server enables leveling, Packet stores activity counts — how many messages you've sent and minutes you've spent in voice there — never the content of anything you say. Those counts power your level, rank, and the server's leaderboards. Your global rank aggregates your counts across the leveling-enabled servers you're in; on global and server-group leaderboards you appear as "Anonymous" unless you opt in to being named — run /level privacy in any server to see exactly what's stored, control your visibility, or delete your progression data entirely. The feature is off by default and counts activity only in servers whose admins enable it.

We keep basic operational records to run and monitor the service: which servers Packet is in and when it was added or removed, and aggregate counts (number of servers, total members, work-queue depth). These describe service operation, not individual people.

To detect and prevent abuse, we record, per request: your IP address and approximate country (from our edge provider), your browser user-agent, the page or route requested, the response status, and timing.

How we store your IP address: as-is, in plain form, for security and abuse-prevention purposes — for example, to recognize and block abusive sources. It is not anonymized or hashed in our records, and it may also appear in our operational logs. (Internally we sign a short-lived token that includes your IP to prevent tampering in transit; that is a transit-only protection and does not change how the IP is stored.)

If an address or a Discord account is blocked for abuse, we keep the blocked value (IP, range, or Discord user ID), the reason, and timing indefinitely as a security record, even after a block is lifted.

We share data only with the infrastructure providers needed to run Packet:

• Discord — your sign-in provider; we call Discord's API with your token to provide the service, and we write the structure you approve back to your server.
• Cloudflare — our security and edge provider. To enforce blocks at the network edge, we send blocked IP addresses/ranges to Cloudflare, and we read back records of requests its protections acted on (which include client IP addresses).
• Our hosting provider — runs the servers and databases.

We do not send any of your data to an AI or large-language-model provider. Packet's "AI Blueprint" feature only gives you a prompt to copy into an AI chat of your own choosing — Packet itself transmits nothing to any AI service.

• Sign out at any time to end your session.
• Download your data — in Settings → Privacy & data, export a copy of the personal data we hold about you.
• Delete your data — in Settings → Privacy & data, permanently delete your personal data: your security/request logs, activity, advanced-call records, feedback, preferences, and your server-role grants.

A few things are handled separately, and we'll tell you plainly:

• Server structure is the server's data, not personal data. Deleting your personal data does not delete a server's blueprints, snapshots, or apply history — other admins of that server may rely on them. Delete those as server data, or they are removed when Packet is removed from the server.
• Abuse records are retained. If your account or IP was blocked for abuse, that record is kept as a security record (so a blocked source can't simply reset it), even after a personal-data deletion.

You can also email us (below) to request a copy or deletion of your data, and we will honour it.

Packet is in active alpha, so this policy may change as the service evolves. We'll update the date at the top when it does.

Questions, or a privacy / data-deletion request? Email [email protected], or use the in-app Feedback page.