The other appendices come from private messages. These are different: they were said publicly, in the bot’s own community server, where anyone could read them. Each one lines up with something the code and data already show — the author just confirmed it in plain words.
Asked whether the bot really keeps everything it’s told — across every server — the operator replied:
“I didn’t add a memory eraser so it just saves like that so stop”
— the operator
What it confirms: the bot has no way to delete what it stores — it just keeps it. That’s the retention behaviour shown in code in the writeup’s Exhibit 02, said out loud.
Pushed to admit what it actually runs on, the bot itself answered with a single word:
“Gemini”
— the bot
What it confirms: the things you say to it are handed to Google’s Gemini, an outside AI model — the third-party pipeline from Exhibit 05. Another person in the chat noted it “still uses gemini.”
On how the bot knows to hand him full control while refusing it to everyone else:
“The scripts knows that I’m the owner by the id”
— the operator
What it confirms: the owner-by-account-ID check from Exhibit 07 — one hard-coded ID unlocks the private console; for everyone else the bot insists it has no owner at all.
On the automatic moderation that watches and acts on the rest of the server:
“btw automod actions do not apply for staff”
— the operator
What it confirms: there really is an enforcement layer making calls on ordinary members — with a quiet exemption for staff. It’s the same watchlist machinery as Exhibit 06.