Responding to an alert¶
A short runbook for a SOC analyst. Every alert record includes the source IP, workstation, user, decoy, protocol and a readable summary.
| Level | What it usually means | What to do |
|---|---|---|
| L3 intent (scan, share listing, login attempts, zone transfer) | Someone is exploring the network. It may be an IT admin, a forgotten scanner, or an intruder in the discovery stage. | Open the host and identify the machine and user. If it is a known scanner, add it to the allow-list. If not, check that machine's EDR/AV and recent logins, and ask its owner. |
| L4 compromise (successful login, file read, command typed) | Someone actively worked on a decoy. Legitimate users have no reason to do this. | Treat it as a probable intrusion. Isolate the source machine (EDR, switch port or VLAN), keep its memory and disk for forensics, and look at what else it touched in your real logs. |
| L5 honeytoken (stolen bait used) | Confirmed malicious: someone stole a fake credential and is using it to move laterally. | Incident response now. Isolate the source, reset the passwords of the users logged on there, and look for the same tools and techniques on other machines. The alert tells you where the bait was taken from: that machine was also visited. |
| L4 sensor offline | A decoy stopped reporting (stopped, isolated or broken). | Check it. An attacker who recognised a decoy may have tried to disable it. For a remote sensor, also check whether its address changed (see Remote sensors). |
| L4 possible DoS | A decoy's connection or thread limits were hit, or a service went down. | Look at the top source addresses named in the incident and treat them like an L3/L4 source. |
Why there are almost no false positives
Nobody has a business reason to use a decoy. The usual benign causes are a vulnerability scanner, an inventory tool or an admin testing a new address. Put those sources on the allow-list once, and every remaining incident deserves a look.