Threat Intelligence and Threat Hunting With MITRE ATT&CK

Defenders who only react to alerts are always a step behind. Threat intelligence gives you knowledge about adversaries (who they are, what they want, how they operate) so that you can prepare. Threat hunting goes further: it is the proactive search for attackers who have already slipped past your automated defences. This lesson explains the concepts, the frameworks, the sources and how to run a simple hunt in your own lab.

What Is Threat Intelligence?

Threat intelligence is evidence-based, contextual and actionable information about existing or emerging threats. Raw data (a list of bad IP addresses) becomes intelligence only when you add context (who uses them, for what, how reliable) and a decision it supports (block, hunt or accept the risk).

Four levels of intelligence

LevelAudienceContentExample
StrategicExecutives, boardLong-term trends, motivations, geopolitical and sector risks“Ransomware groups increasingly target regional hospitals”
OperationalSecurity managers, incident commandersCampaigns, adversary intent and timing“A campaign using fake invoice emails is hitting our sector this month”
TacticalSOC, hunters, engineersTactics, techniques and procedures (TTPs)“The group uses scheduled tasks for persistence”
TechnicalTools and analystsIndicators of compromise (IOCs): hashes, domains, IPs, file names“Block this domain, search for this file hash”

Indicators: IOCs vs TTPs

Security teams talk about the Pyramid of Pain, a model by David Bianco that ranks indicators by how hard it is for the attacker to change them:

Indicator typePain to attacker if you detect it
Hash valuesTrivial (change a byte)
IP addressesEasy
Domain namesSimple
Network and host artefactsAnnoying
ToolsChallenging
Tactics, techniques and procedures (TTPs)Tough: they must change how they operate

The lesson: use IOCs for quick wins, but build detections around behaviour (TTPs) for lasting value.

Frameworks and Languages

FrameworkUse
MITRE ATT&CKKnowledge base of adversary tactics and techniques; the Enterprise matrix has 14 tactics from Reconnaissance to Impact
Lockheed Martin Cyber Kill ChainSeven-stage model of an intrusion; see Threat Actors and the Attack Lifecycle
Diamond ModelRelates adversary, capability, infrastructure and victim
STIX / TAXIIStandard format and transport protocol for sharing intelligence
SigmaGeneric detection rule format convertible to many SIEM languages
YARARules that identify files by patterns; see Malware Types and Analysis Basics
TLP (Traffic Light Protocol)Labels (RED, AMBER, GREEN, CLEAR) that control how intelligence may be shared

Intelligence Sources

  • Open source (OSINT): vendor blogs, government advisories (CISA, CERT-In, NCSC), research reports, social media.
  • Community and sharing groups: ISACs for sectors such as finance, health and energy; trusted peer groups.
  • Commercial feeds and platforms: paid intelligence and managed services.
  • Open platforms: MISP (open-source threat sharing platform), AlienVault OTX, abuse.ch projects, VirusTotal for file and URL reputation.
  • Internal intelligence: your own incidents, logs, honeypots and phishing reports, often the most relevant of all.

Evaluating a source: the Admiralty-style questions

  • How reliable has this source been before?
  • Is the information corroborated by independent sources?
  • How old is it? Infrastructure indicators decay quickly.
  • Is it relevant to my organisation, sector, region and technology?
  • Can I act on it?

The Intelligence Cycle

  1. Direction: decide what questions matter (“Which groups target our sector?”).
  2. Collection: gather data from sources.
  3. Processing: normalise and de-duplicate.
  4. Analysis: add context, judge confidence and draw conclusions.
  5. Dissemination: deliver to the right audience in the right format.
  6. Feedback: refine requirements based on what was useful.

Threat Hunting

Hunting assumes that a skilled adversary may already be inside. It is a human-led, hypothesis-driven investigation that complements automated detection.

ApproachStarting pointExample
Hypothesis-drivenA theory based on intelligence or ATT&CK“Attackers may be using scheduled tasks to persist; do we see unusual tasks?”
Intelligence-drivenNew IOCs or a campaign reportSearch the last 90 days of DNS logs for a newly reported domain
Data-driven / anomalyStatistics and baselinesFind rare parent-child process pairs or unusual login hours

The hunt loop

  1. Hypothesis: write it down, including the ATT&CK technique.
  2. Data: list the log sources needed and verify they exist and are searchable.
  3. Query and analyse: search, pivot, visualise and test.
  4. Result: found activity (escalate to incident response), found nothing (record the coverage), or found a visibility gap (fix logging).
  5. Automate: turn any good hunt into a standing detection rule.
  6. Document: share the hunt so that others can repeat it.

Practice Lab 1: Research a Real Technique With ATT&CK

  1. Open the MITRE ATT&CK website (attack.mitre.org) and find technique T1053 (Scheduled Task/Job).
  2. Read the description, procedure examples, detection and mitigation sections.
  3. List the log sources and events you would need to detect it on Windows and Linux.
  4. Write your hunt hypothesis in one sentence.

Practice Lab 2: Run the Hunt in Your Lab

On a lab Linux VM, create a baseline and an unexpected cron entry, then hunt for it:

# Baseline: record normal scheduled jobs
sudo ls -l /etc/cron.* /var/spool/cron/crontabs 2>/dev/null > baseline_cron.txt
crontab -l 2>/dev/null >> baseline_cron.txt

# Simulate an unexpected persistence mechanism (harmless)
( crontab -l 2>/dev/null; echo "*/30 * * * * /usr/bin/logger lab-hunt-test" ) | crontab -

# Hunt: compare to baseline
sudo ls -l /etc/cron.* /var/spool/cron/crontabs 2>/dev/null > current_cron.txt
crontab -l >> current_cron.txt
diff baseline_cron.txt current_cron.txt

# Check what logged the change
sudo grep CRON /var/log/syslog | tail
sudo grep -i "crontab" /var/log/auth.log | tail

Remove the test entry afterwards with crontab -e. On a Windows lab VM, do the equivalent by listing scheduled tasks with Get-ScheduledTask before and after creating a harmless task and by finding event ID 4698 (task created) in the Security log, if auditing is enabled.

Practice Lab 3: Use IOCs Defensively

  1. Pick a recent public advisory from CISA or CERT-In that lists indicators.
  2. Extract five IOCs (domains or hashes). Check their age and reputation using VirusTotal’s website.
  3. Write a small script that searches your lab logs for those strings:
    grep -R -n -F -f iocs.txt /var/log 2>/dev/null | head
  4. Explain why a clean result tells you less than it appears to.

Practice Lab 4: Write a Sigma Rule

title: New Scheduled Task Created via Command Line
status: experimental
description: Detects creation of scheduled tasks using schtasks, a common persistence technique
logsource:
  category: process_creation
  product: windows
detection:
  selection:
    Image|endswith: '\schtasks.exe'
    CommandLine|contains: '/create'
  condition: selection
falsepositives:
  - Legitimate administration and software installers
level: medium
tags:
  - attack.persistence
  - attack.t1053.005

Think about the false positives. How would you reduce them (known administrators, known paths, time windows)? Convert the rule for your SIEM using the Sigma tooling (sigma-cli) and test it against lab events.

Common Mistakes

  • Collecting feeds without a purpose, and blocking indicators that are stale or shared infrastructure.
  • Treating hunting as running random searches without a hypothesis.
  • Hunting with no logs, so nothing can be found.
  • Not turning successful hunts into permanent detections.
  • Ignoring sharing rules such as TLP.

Practice Exercises

  1. Explain the difference between an IOC and a TTP, with one example of each.
  2. Write three hunt hypotheses based on different ATT&CK tactics.
  3. Describe how you would evaluate the reliability of a new threat feed.
  4. Where would the Pyramid of Pain direct your detection effort for ransomware?
  5. Define a requirement for a strategic intelligence report for a hospital board.

Mentor tip: A hunt that finds nothing is not a failure if you recorded what you looked at and what data you had. Over time that record shows exactly where you are blind, which is valuable.

Educational use only: hunt and test only on systems you own or are authorised to monitor.

Key takeaways
  • Threat intelligence turns data into context and decisions, at strategic, operational, tactical and technical levels.
  • Behaviour-based detection (TTPs) outlasts indicator-based blocking (IOCs).
  • MITRE ATT&CK, STIX/TAXII, Sigma, YARA and TLP are the common tools and conventions.
  • Threat hunting is hypothesis-driven and complements automated detection.
  • Convert successful hunts into standing detections.

Related Lessons