Skip to content

Incidents and levels

Counting events is useless: a single nmap run produces hundreds. ReD-Pot scores how far a source got instead.

The severity ladder

Level Name Meaning Examples Alert
L1 touch TCP handshake only someone typed a wrong IP no (logged)
L2 interaction protocol data exchanged HTTP GET, SMB negotiate, SSH banner, DNS lookup, LDAP rootDSE no (logged)
L3 intent trying to get in or map things login attempt, share enumeration, zone transfer, Kerberos pre-auth guess, port scan, password spraying, path brute force yes
L4 compromise it worked valid login, file read, command typed, sensitive file downloaded, upload attempt, decoy offline (tampering), a bait account name used without its secret yes
L5 honeytoken stolen bait used a planted password, SSH key or document used anywhere with the correct secret, a TGT for a bait account, a Kerberoast of the honey SPN yes, critical

Escalation rules:

  • Port scan: 5 or more distinct decoy ports in 10 minutes, or every port of one decoy with at least 3 ports.
  • Password spraying: 5 or more failures in 5 minutes.
  • Path brute force on the web decoy: many 404s in a short time.
  • Connection floods and resource exhaustion raise their own events; a saturated decoy opens an L4 "possible DoS" incident that names the top source addresses.

Generic user names such as admin, administrator, root or backup are in every wordlist, so on their own they never count as a honeytoken. A matching bait password always does.

Incidents

Events from the same source are grouped into one incident, which only ever goes up the ladder. An incident shows the source (IP, workstation and logged-on user when known), the decoys it touched, the time window, the level and the full timeline.

Two kinds of incident exist:

Kind Source Raised by
attacker a source IP events from that source
decoy health the decoy itself (decoy FS01) a decoy that stopped reporting (L4 "possible tampering") or a service that went down or was saturated (L4 "possible DoS")

Allow-listed sources (your vulnerability scanner, IT admin hosts) are recorded but never raise incidents. See Configuration file.

Status

Open an incident to change its status:

Status Use it when
Open new, not looked at yet
Acknowledged someone is working on it
Closed handled
False positive the activity was expected (for example a forgotten scanner); consider adding the source to the allow-list

Privacy of tried passwords

Passwords that people try on a decoy are stored masked, for example S******* (len 9, h:1a2b3c4d), so an employee who mistypes a real password into a decoy does not end up in your SIEM in clear text.