// THE SWARM

Six agents. One per attack class.

Each agent goes deep on a single class of vulnerability, and they share context as they work — so findings that span files get chained, not missed.

Follows every untrusted input across files to the sink it reaches.

Sweeps source and git history for what should never be committed.

Maps your dependency tree to CVEs that are actually reachable.

Checks every route and object reference for missing authorization.

Probes order-of-operations flaws: races, replays, and steps that can be skipped.

Catches misconfiguration in code, before it reaches production.