Intent and Data Implementation

1. Requested Intents

V2 starts without privileged intents unless the matching environment setting is explicitly enabled after approval in the Discord Developer Portal. This prevents accidental use of disallowed intents.

const intents = [GatewayIntentBits.Guilds, GatewayIntentBits.GuildMessages];

if (messageContentApproved) intents.push(GatewayIntentBits.MessageContent);
if (guildMembersApproved) intents.push(GatewayIntentBits.GuildMembers);

2. V1 Server Members Intent

The application reacts to member join and leave events. It records a limited event history and uses server configuration to send verification instructions. Role assignment and removal are performed only for configured ThanHub verification and premium workflows.

client.on("guildMemberAdd", async member => {
  await memberHistory.insert({
    guildId: member.guild.id,
    userId: member.user.id,
    username: member.user.username,
    eventType: "join"
  });

  const verification = await getVerificationConfig(member.guild.id);
  if (verification) await sendVerificationInstructions(member, verification);
});

client.on("guildMemberRemove", async member => {
  await memberHistory.insert({
    guildId: member.guild.id,
    userId: member.user.id,
    username: member.user.username,
    eventType: "leave"
  });
});

3. V1 Message Content Intent

V1 needs message content for existing prefix commands beginning with th. Ordinary messages increment numerical daily, weekly, and total counters; their text is not written to the activity-count table.

client.on("messageCreate", async message => {
  if (message.author.bot || !message.guild) return;

  await incrementNumericActivityCounts(message.author.id, message.guild.id);

  if (!message.content.toLowerCase().startsWith("th")) return;
  const [commandName, ...args] = parsePrefixCommand(message.content);
  await runApprovedCommunityCommand(commandName, args);
});

• Processed transiently: command name and arguments needed to execute the requested command.
• Stored: user ID, guild ID, daily count, weekly count, total count, and reset dates.
• Not stored by activity tracking: ordinary message bodies.
• User choice: members can avoid prefix parsing by not invoking th commands and using slash commands where available.

4. V2 Private Ticket Processing

V2 immediately ignores bots, DMs, and non-ticket channels. Purchase and giveaway tickets are manual. Support-ticket AI remains disabled until the ticket owner selects Request AI Help.

async function handleTicketMessage(message) {
  if (!message.guild || message.author.bot || !isTicketChannel(message.channel)) {
    return false;
  }
  if (isPurchaseTicket(message.channel) || isGiveawayTicket(message.channel)) {
    return true; // Manual staff workflow only
  }
  if (ticketIsClosed(message.channel) || !aiHelpWasRequested(message.channel)) {
    return true;
  }
  queueSupportReply(message);
  return true;
}

Casual or insufficient messages are filtered before any Gemini request. Relevant support text is sent only for immediate inference and is not used by ThanHub to train or fine-tune models.

5. V2 Scam-Trap Moderation

The safety handler processes messages only when they are posted in one specifically configured scam-trap channel. Bot messages and members with the configured exemption role are ignored.

async function handleScamTrapMessage(message) {
  if (!message.guild || message.author.bot || message.channelId !== SCAM_TRAP_CHANNEL) {
    return false;
  }
  if (memberHasExemptRole(message.member)) return false;

  await message.delete();
  if (message.member.moderatable) {
    await message.member.timeout(CONFIGURED_DURATION, MODERATION_REASON);
  }
  await notifyUserAndWriteModerationLog(message);
  return true;
}

• The triggering message is deleted from Discord.
• Moderation metadata may be retained: affected user, channel, reason, timestamp, and action result.
• The warning panel tells members not to post in the protected channel.

6. Ticket Transcript Storage

When authorized staff closes a ticket, the bot gathers ticket messages, creates a random unguessable token, and sends the transcript to a protected API authenticated by a server-side secret. The secret is never exposed in Discord or browser code.

Stored transcript records may include ticket owner ID, channel and guild IDs, channel name, staff claimer, closer, timestamps, message text, and attachment links. Access is limited to authorized ThanHub staff and protected infrastructure.

7. Retention, Security, and Deletion

• Data is retained only for community operations, licensing, security, support, moderation, and accountability.
• Production records use access-controlled Supabase and ThanHub infrastructure; credentials remain server-side.
• Discord API data is not sold, used for advertising, or used to train ThanHub AI models.
• Eligible access, correction, or deletion requests can be submitted through the ThanHub support server with the requester's Discord user ID.
• Security, moderation, or transaction records may be retained where needed to prevent abuse or meet operational obligations.

8. Reviewer Evidence Checklist

The following in-server demonstrations correspond directly to the implementation above:

• V1 member joins and receives configured verification instructions.
• V1 member leaves and a limited join/leave event is recorded.
• V1 processes a visible th prefix command and updates numerical activity statistics.
• V2 creates a private support ticket where AI is initially disabled.
• The ticket owner explicitly enables AI help before a support response is generated.
• Authorized staff closes a ticket and a private transcript is generated.
• A non-exempt member posts in the clearly labeled scam-trap channel and receives the configured moderation action.

Related public documents: Bot Data Usage Policy, Privacy Policy, and Terms of Service.