Argus
It's already watching.
What it watches for
19+ rules. One purpose.
A hand-written detection engine, not a keyword filter: raid response, credential leaks, deceptive links, impersonation, webhook abuse, and more, each evaluated on every message and every audit-log entry in real time.
CREDENTIAL LEAK
RAID RESPONSE
NUKE / TAMPERING
SCAM LINK
QR CODE THREAT
IP-GRABBER LINK
STAFF IMPERSONATION
WEBHOOK ABUSE
The machine
Built to not miss anything
How it thinks
Real engineering, not a feature list
It watches itself, too
An audit-log entry that targets Argus's own role or permissions fires a self-defense rule ; disabling the security bot is step one of any competent attack, and it used to be completely silent.
Nothing gets silently dropped
Every action Argus takes is written to an AES-256 encrypted audit log (SQLCipher at rest, 30-day retention), searchable by an admin, with flagged content automatically stripped of credentials before it's ever stored.
A global threat picture
Every link and IP is checked against threat intelligence shared across every server Argus protects, not built from just this one's own history: scam domains, malicious netblocks, and known-bad IPs, refreshed continuously from four independent sources.
Independence, not exclusivity
Mandatory CAPTCHA and the admin trust-score watchlist are two fully separate systems on purpose: a low-trust joiner can be flagged AND still have to solve the challenge, not one or the other.
A raid can't out-patiently wait
Creation-time clustering checks whether recent joiners' Discord accounts were all registered within minutes of each other ; aging an account or trickling joins alone doesn't beat it, since it costs zero API calls to check.
Argus
Standalone Discord security.
Built quietly. Watching already.
Argus is a separate project from .bOne: its own bot, its own engine, still actively being built and tested. It isn't installable yet.