When you add a security bot to your community you hand it the audit log, the messages and sometimes Administrator. In exchange, you trust whoever wrote it. There is no mechanism: it is assumed. This page exists so that here nothing has to be assumed.
None. And the day this stops being zero, this number will say so all the same.
/support allow, and only the server owner can. It expires on its own (30 minutes by default, four hours at most) and you close it whenever you want with /support revoke./support mode to "never" and we never get in, not even in an emergency. The price is stated where you turn it on: if something breaks in there, we will not be able to look at it./support history shows you all of it, and /integrity checks the log has not been touched.tools/verify.py on it — it depends on nothing of ours. It checks the signatures, the chain, and that the log has never shrunk.This is not end-to-end encryption. Your data sits in the clear on our disk and the bot process reads it, because moderating means seeing what you moderate; anyone with root on the machine would open the database and no door stops that. What this bounds is the normal path — the panel, which is where looking actually happens — and what it guarantees is not "it is impossible", it is "it cannot be done without it being seen". We would rather say it that way than promise more.