HTB CDSA study materials
HTB CDSA study resources

HTB CDSA cheat sheet

HTB CDSA · Keep the key ideas within reach.

Quick reference

Incident Response and Threat Hunting

  • Incident response lifecycle per NIST SP 800-61: Preparation, Detection and Analysis, Containment/Eradication/Recovery, Post-Incident Activity.

  • Severity triage inputs: data sensitivity, blast radius (hosts/accounts affected), attacker dwell time, and whether persistence or lateral movement is confirmed.

  • Order of volatility for collection: CPU registers/cache, RAM, network state, running processes, disk, remote logs, archival backups.

  • Hunting hypothesis structure: threat actor TTP (from ATT&CK) + observable data source + expected artifact; a hunt without a data source is not executable.

  • Containment decision rule: network isolation before credential reset when active C2 is suspected; credential reset first when only credential misuse is observed.

  • Dwell time = first malicious artifact timestamp to detection timestamp; measured from earliest reliable log, not from alert time.

  • Eradication is verified by absence of the persistence mechanism across all scoped hosts, not by absence of the original alert.

  • Lessons-learned output must map each finding to a detection gap (missing log source, missing rule, or missing telemetry field).

Quick reference

SIEM and Log Analysis

  • Windows Security event IDs: 4624 logon, 4625 failed logon, 4634/4647 logoff, 4648 explicit credentials, 4672 special privileges, 4688 process creation, 4720 account created, 4768/4769/4771 Kerberos TGT/TGS/preauth failure.

  • Logon type codes: 2 interactive, 3 network, 4 batch, 5 service, 7 unlock, 8 network cleartext, 9 new credentials, 10 RDP, 11 cached interactive.

  • Sysmon IDs: 1 process create, 3 network connect, 7 image load, 8 CreateRemoteThread, 10 process access, 11 file create, 12/13/14 registry, 22 DNS query, 25 process tampering.

  • Correlation rule anatomy: aggregation key (user, host, source IP), threshold, time window, and suppression to prevent alert storms.

  • EPSS gives probability of exploitation in 30 days; CVSS base score measures severity, not likelihood; prioritization uses both plus asset criticality.

  • Clock skew breaks correlation: normalize all sources to UTC and alert when a source deviates beyond an agreed tolerance.

  • Sigma rules are vendor-neutral detections compiled to SIEM query languages; the logsource block (product, service, category) determines applicability.

  • Absence-of-event detections (e.g., no 4688 from a host for 24h) often indicate agent tampering rather than a quiet endpoint.

Quick reference

Network Traffic Analysis and Intrusion Detection

  • TCP flag interpretation: SYN only = connection attempt; SYN/ACK = open port; RST = closed or filtered; FIN/PSH/URG combination is used by nmap NULL/FIN/XMAS scans.

  • Beaconing detection: low variance in connection intervals to a single destination, consistent payload size, and periodic jitter are stronger signals than destination reputation.

  • DNS tunneling indicators: high volume of TXT/NULL queries, long encoded subdomain labels, high entropy in query names, and one authoritative domain receiving all queries.

  • Suricata/Snort rule anatomy: action, protocol, source/destination, direction operator (-> or <>), rule options in parentheses, sid and rev for versioning.

  • PCAP triage order: protocol hierarchy statistics, conversations sorted by bytes, then follow-stream on anomalous flows.

  • JA3/JA3S fingerprints hash TLS client/server hello parameters; a known-malware JA3 is a durable detection even when domains rotate.

  • SMB lateral movement indicators: 445/tcp sessions from workstations to workstations, ADMIN$ or IPC$ access, and service creation via SVCCTL.

  • IDS evasion: fragmentation overlap, TCP segmentation, and protocol ambiguity require stream reassembly enabled on the sensor to be detected.

Quick reference

EDR and Host-Based Analysis

  • Process tree anomaly rules: Office or browser spawning cmd/powershell/wscript; lsass access by non-security processes; svchost with a non-system parent.

  • Command-line hunting strings: -enc/-EncodedCommand, -nop, -w hidden, IEX, DownloadString, FromBase64String, certutil -urlcache, bitsadmin /transfer.

  • LSASS access detection: Sysmon Event ID 10 with GrantedAccess 0x1010 or 0x1410 and a calling process outside known security tooling.

  • Persistence locations to check: Run/RunOnce keys, services, scheduled tasks, WMI event subscriptions, startup folders, and logon scripts.

  • Parent PID reuse and orphaned processes: a process whose parent no longer exists may indicate injection or a terminated loader.

  • Memory analysis order: process list, network connections, injected memory regions (RWX), loaded modules, then strings and YARA scans.

  • Living-off-the-land binaries (LOLBins): certutil, regsvr32, mshta, rundll32, wmic, and installutil are signed and abused for download or execution.

  • EDR isolation blocks network but preserves agent channel; it does not stop local process execution, so memory capture should precede or accompany it.

Quick reference

Digital Forensics and Malware Analysis

  • NTFS artifacts: $MFT (file records), $UsnJrnl:$J (change journal), $LogFile (metadata transactions), $Secure, and alternate data streams.

  • Windows timestamps: $STANDARD_INFORMATION can be modified by timestomping; $FILE_NAME timestamps in the MFT are harder to alter and are compared for discrepancy.

  • Registry hives: SYSTEM (services, timezone), SOFTWARE (installed apps, run keys), SAM (local accounts), SECURITY (cached credentials), NTUSER.DAT (per-user activity).

  • Prefetch shows execution count and last run; Amcache and ShimCache indicate presence/execution; BAM/DAM records background activity per user SID.

  • Static malware analysis order: hashes, strings, PE headers and sections, imports, entropy per section, then YARA matching.

  • High section entropy plus a small import table suggests packing; unpack by running in an instrumented sandbox and dumping memory.

  • Chain of custody requires hash of the image at acquisition, documented handler, and hash verification at each transfer.

  • Anti-forensics indicators: event log clearing (1102), USN journal deletion, timestomping, and secure-deletion tools leaving sparse MFT records.

Quick reference

Active Directory and Windows Security Monitoring

  • Kerberoasting: 4769 with RC4 encryption type (0x17) for service accounts, often high volume from one source; defense is AES-only and long service-account passwords.

  • AS-REP roasting: 4768 with pre-authentication disabled (0x0) returned for accounts lacking 'Do not require Kerberos preauthentication'.

  • DCSync: replication rights abuse detected via 4662 with GUID 1131f6ad-9c07-11d1-f79f-00c04fc2dcd2 (DS-Replication-Get-Changes) from a non-DC principal.

  • Golden ticket: forged TGT with abnormal ticket lifetime and a domain SID from a krbtgt hash; validate with 4769 anomalies and krbtgt double-reset.

  • Pass-the-hash indicators: 4624 type 3 with NTLM authentication, no preceding 4648, and mismatched workstation name versus source IP.

  • Domain controller audit policy must enable 'Audit Directory Service Access' for 4662 and 'Audit Kerberos Authentication Service' for 4768/4769.

  • BloodHound-style attack paths rely on ACL edges (GenericAll, WriteDACL, ForceChangePassword); monitor 5136 directory changes for ACL modification.

  • GPO abuse: modification of SYSVOL scripts or new scheduled tasks pushed domain-wide appears as 5136 on groupPolicyContainer objects and 4688 on many hosts.

Quick reference

Reporting and Communication

  • Executive summary states business impact, current status, and decisions needed in the first paragraph; technical detail follows in appendices.

  • Findings must be reproducible: include the query, log source, time range, and host or account identifiers for every claim.

  • Timeline entries require a timestamp, source system, actor, action, and confidence level; unverified entries are labeled as hypotheses.

  • IOC table columns: type, value, first seen, last seen, confidence, and recommended action (block, hunt, monitor).

  • Severity language aligns to a defined scale (e.g., Critical/High/Medium/Low) with explicit criteria so readers cannot reinterpret labels.

  • ATT&CK mapping in reports shows technique ID, observed evidence, and detection coverage status, enabling gap analysis.

  • Recommendations are prioritized by risk reduction per effort and assigned an owner and due date, otherwise they are not actionable.

  • Regulatory and contractual notification triggers (e.g., personal data exposure) are decided with legal, and the decision is documented with rationale.