Identity Compromise Investigation

Accounts are the most common path to impact. Decide whether an account is compromised, how far it was used, and what containment is justified.

Key events and the Wazuh rules that surface them

Rule numbers are from the stock Wazuh 4.14 ruleset. Many Windows events are only alerted on when the channel is collected and not filtered. If there's no rule, search the raw event with search_security_events (field data.win.system.eventID). Load wazuh-windows-sysmon for Windows field paths.

Platform	Event	Wazuh rules	Meaning
Windows	Event ID 4624	rule 60106 (level 3)	successful logon; check data.win.eventdata.logonType: 2 interactive, 3 network, 10 RemoteInteractive/RDP
Windows	Event ID 4625	rules 60105, 60122 (level 5); burst → rule 60204 (level 10)	failed logon; subStatus separates bad password from unknown user
Windows	Event ID 4648	none (search raw events)	logon with explicit credentials (runas, lateral tools)
Windows	Event ID 4672	rule 67028 (level 3)	special privileges assigned: admin-equivalent logon. Low level, high meaning for non-admins
Windows	Event ID 4720, Event ID 4722	rule 60109 (level 8)	account created / enabled
Windows	Event ID 4738	rule 60110 (level 8)	account changed
Windows	Event ID 4728, Event ID 4732, Event ID 4756	rules 60141, 60144, 60151 (level 5); rule 60154 Administrators group changed (level 12)	member added to security-enabled global / local / universal group
Windows	Event ID 4740 lockout	rule 60115 (level 9)	account locked out after multiple failures
Windows	Event ID 4768, Event ID 4771	none stock (search raw events)	Kerberos TGT request / pre-auth failure: spraying against a DC
Linux	sshd failure	rules 5716, 5760 (level 5); rule 5710 non-existent user (level 5)	single failures
Linux	sshd burst	rules 5712, 5720, 5763 (level 10); rule 5758 max attempts exceeded (level 8)	composite brute-force rules
Linux	sshd success	rule 5715 (level 3)	Accepted password / publickey
Linux	sudo	rule 5402 to root (level 3), rule 5401 failed (level 5), rule 5403 first time for user (level 4)	privilege use
Linux	PAM	rule 5503 login failed (level 5), rule 5551 multiple failures (level 10)	console / other PAM services
Linux	account changes	rule 5902 new user (level 8), rule 5904 user information changed (level 8)	useradd / usermod

Procedure

1. List all auth events for the user over 7 days (search_security_events(query="<user>")), and aggregate by source IP and host (get_alerts_aggregated).
2. Establish baseline: usual source IPs, hosts, hours, logon types.
3. Test each hypothesis below; record which are supported.
4. Determine scope: every host the account touched after the suspected compromise time.

Hypotheses and indicators

Pattern	Indicators	ATT&CK
Brute force → success	rule 60204 or rules 5712, 5763 then rule 60106 or rule 5715 from the same source	T1110 → T1078
Password spraying	one source, many users, few attempts each	T1110.003
Impossible travel	successful logons from locations whose distance/time implies > ~900 km/h	T1078
New source / off-hours	success from never-seen IP/ASN or outside baseline hours	T1078
Privilege escalation	rule 67028 for a non-admin, group adds (rules 60141, 60144, 60154), rule 5402 by an unusual user	T1078.002/.003, T1098
Lateral movement	same account, new hosts in short time; logon type 3/10 chains; SSH hops	T1021 (.001 RDP, .002 SMB, .004 SSH)
Persistence via account	new account created by the suspect account (rules 60109, 5902), SSH key added (FIM on authorized_keys, see wazuh-fim-investigation)	T1136, T1098.004

Verdict

• Compromised — success from attacker-controlled source, or malicious actions by the account. Confidence ≥ 0.8.
• Targeted, not compromised — failures only, no success. Recommend blocking the source, not disabling the user.
• Inconclusive — anomalies without malicious actions; recommend user verification.

Recommend disable_user (proposed by response-planner) only with evidence of successful malicious use. Service accounts and break-glass accounts: flag the business impact explicitly; prefer blocking the source or isolating the host first.

Output

• add_entities (user as victim; source IPs as attacker; hosts touched).
• link_mitre for supported techniques.
• update_case severity/confidence.
• add_finding titled Identity assessment: <user> with baseline, hypotheses table (supported / not supported + evidence), hosts in scope, verdict, recommended containment. standard_refs: NIST-800-61r3, CIS-v8.1:5 (Account Management), CIS-v8.1:6 (Access Control Management).