Skip to content

MITRE ATT&CK for ICS: A Practical Guide

Use MITRE ATT&CK for ICS in practice: its 12 tactics, sub-techniques, assets and detections, with jq queries, Navigator layers and coverage mapping for OT.

Published on 8 min read

MITRE ATT&CK for ICS is a knowledge base of adversary behavior against industrial control systems, built from public reporting on real intrusions. It is the OT counterpart of the better known Enterprise matrix, and it gives engineers, defenders and threat intelligence teams a shared vocabulary for questions such as "how would an attacker stop operators from seeing the process?" or "which of our detections would catch an unauthorized program download?".

This guide explains how the matrix is organized today, what changed recently, and how to use it for concrete work: threat modeling, detection coverage, incident reporting and exercise planning.

How the matrix is organized

The ICS matrix has 12 tactics, the adversary's goals at each stage:

IDTacticWhat it covers
TA0108Initial AccessGetting a foothold in the ICS environment
TA0104ExecutionRunning code or commands, including on controllers
TA0110PersistenceKeeping access across restarts and changes
TA0111Privilege EscalationGaining higher permissions
TA0103EvasionAvoiding detection by tools and operators
TA0102DiscoveryLearning the network and the devices on it
TA0109Lateral MovementMoving between systems and zones
TA0100CollectionGathering process and configuration information
TA0101Command and ControlCommunicating with compromised systems
TA0107Inhibit Response FunctionPreventing safety, protection and operator response
TA0106Impair Process ControlManipulating the physical process
TA0105ImpactThe effect on the process, people, equipment or business

The last three tactics are what make the ICS matrix different. Inhibit Response Function covers actions such as Alarm Suppression (T0878), Modify Alarm Settings (T0838) and Device Restart/Shutdown (T0816). Impair Process Control covers Modify Parameter (T0836), Brute Force I/O (T0806) and unauthorized command messages. Impact uses process language rather than IT language: Loss of Safety (T0880), Loss of View (T0829), Denial of View (T0815), Manipulation of Control (T0831), Damage to Property (T0879).

Sub-techniques arrived in ICS

For years the ICS matrix had no sub-techniques. Recent releases changed that, and the current version (v20 at the time of writing) groups related behaviors under parent techniques. Several older technique IDs were consolidated, and their pages now redirect:

Old IDCurrent IDName
T0855 Unauthorized Command MessageT1692.001Unauthorized Message: Command Message
T0856 Spoof Reporting MessageT1692.002Unauthorized Message: Reporting Message
T0857 System FirmwareT1693.001Modify Firmware: System Firmware
T0812 Default CredentialsT1694.001Insecure Credentials: Default Credentials
T0891 Hardcoded CredentialsT1694.002Insecure Credentials: Hardcoded Credentials
T0803 Block Command MessageT1691.001Block Operational Technology Message: Command Message

Some existing techniques gained sub-techniques while keeping their ID, for example Program Download (T0843) now has Download All (T0843.001), Online Edit (T0843.002) and Program Append (T0843.003), and Remote System Discovery (T0846) has Port Scan (T0846.001) and Broadcast Discovery (T0846.002) among others. If your detection rules, reports or SIEM tags reference old IDs, plan a migration and keep a mapping table.

Assets, software, groups and campaigns

Beyond tactics and techniques, the ICS knowledge base documents:

  • Assets: the device classes techniques apply to, such as Programmable Logic Controller (A0003), HMI, engineering workstation, safety controller and data historian.
  • Software: ICS specific malware such as Stuxnet (S0603), Industroyer (S0604), Industroyer2 (S1072), Triton (S1009) and INCONTROLLER (S1045).
  • Groups: for example Sandworm Team (G0034) and TEMP.Veles (G0088).
  • Campaigns: for example the 2015 Ukraine Electric Power Attack (C0028), the 2016 Ukraine Electric Power Attack (C0025) and the Triton Safety Instrumented System Attack (C0030). The ICS attack case studies summarize these events.
  • Mitigations: ICS specific mitigations such as Network Segmentation (M0930), Filter Network Traffic (M0937), Authorization Enforcement (M0800) and Software Process and Device Authentication (M0813).
  • Detection strategies and analytics: recent versions describe detection as detection strategies with platform specific analytics, linked to data components (for example network traffic content and flows, device alarms, process history).

Getting the data

Working with ATT&CK programmatically is easier than reading the website. MITRE publishes the full knowledge base as STIX 2.1 JSON in the mitre-attack/attack-stix-data repository, with one bundle per domain:

curl -sLo ics-attack.json \
  https://raw.githubusercontent.com/mitre-attack/attack-stix-data/master/ics-attack/ics-attack.json

List every current technique with its tactic, skipping revoked and deprecated entries:

jq -r '.objects[]
  | select(.type=="attack-pattern" and (.revoked|not) and (.x_mitre_deprecated|not))
  | [(.external_references[] | select(.source_name=="mitre-attack").external_id),
     .name,
     ([.kill_chain_phases[].phase_name] | join(","))]
  | @tsv' ics-attack.json | sort

List the techniques attributed to a piece of software, here Triton (S1009):

jq -r '
  (.objects | map(select(.type=="malware"))
            | map(select(any(.external_references[]; .external_id=="S1009")))
            | .[0].id) as $sw
  | (.objects | map(select(.type=="attack-pattern"))
              | map({key: .id, value: ((.external_references[]
                     | select(.source_name=="mitre-attack").external_id) + "  " + .name)})
              | from_entries) as $tech
  | .objects[]
  | select(.type=="relationship" and .relationship_type=="uses" and .source_ref==$sw)
  | $tech[.target_ref] // empty' ics-attack.json | sort

The output includes Program Download (T0843), Modify Controller Tasking (T0821), Change Operating Mode (T0858), Detect Operating Mode (T0868), System Firmware (T1693.001) and Loss of Safety (T0880), which matches the public analysis of the malware's interaction with safety controllers. For Python users, MITRE also maintains the mitreattack-python library, listed on the ATT&CK data and tools page.

Use case 1: threat modeling a zone

Pick one zone, for example a packaging line with two PLCs, an HMI and an engineering workstation. Walk the tactics from left to right and ask which techniques are feasible given your architecture:

  • Initial Access: is there External Remote Services (T0822) into this zone? Transient Cyber Asset (T0864), meaning vendor laptops and USB drives? Internet Accessible Device (T0883)?
  • Lateral Movement: can the engineering workstation reach the PLCs with engineering protocols? It must, so Program Download (T0843) is feasible from that host.
  • Inhibit Response Function and Impair Process Control: which safety and protection functions does this zone host, and could Modify Parameter (T0836) on these PLCs cause harm?

The result is a short list of plausible techniques per zone that feeds segmentation decisions and IEC 62443 zone risk assessments (see IEC 62443 explained).

Use case 2: detection coverage

ATT&CK is most useful when you map what you can actually see. For each relevant technique, record the data source you collect and the detection you run. A passive OT network sensor (see OT asset inventory with passive monitoring) covers many of the high value ones:

TechniqueObservableExample detection
Program Download (T0843)Engineering protocol download function codesAlert on downloads outside a change window or from a host that is not an approved engineering workstation
Change Operating Mode (T0858)Mode change commands, controller diagnostic bufferAlert on run to program or stop transitions
Remote System Discovery (T0846)Scans and broadcast identity requestsAlert on new hosts sending discovery traffic in Level 1 and 2
Unauthorized Message: Command Message (T1692.001)Write requests from unexpected sourcesAllow list of which hosts may write to which controllers
Remote Services (T0886)RDP, VNC, SSH into OTAlert on sessions not originating from the jump host
Alarm Suppression (T0878)Alarm and event logs, HMI configurationAlert on alarm disable or shelving outside procedure

Then load your coverage into the ATT&CK Navigator to see gaps. A minimal Navigator layer for the ICS domain:

{
  "name": "Packaging line detection coverage",
  "domain": "ics-attack",
  "versions": { "attack": "20", "layer": "4.5" },
  "techniques": [
    { "techniqueID": "T0843", "score": 2, "comment": "Sensor alert on download outside change window" },
    { "techniqueID": "T0858", "score": 1, "comment": "Mode changes logged, no alert yet" },
    { "techniqueID": "T0886", "score": 2, "comment": "RDP not from jump host alerts" },
    { "techniqueID": "T0878", "score": 0, "comment": "No visibility" }
  ],
  "gradient": { "colors": ["#ff6666", "#ffe766", "#8ec843"], "minValue": 0, "maxValue": 2 }
}

Open the Navigator, choose "Open Existing Layer", and upload the file. Be honest with scores: "we log it somewhere" is not the same as "an analyst would see it in time".

Use case 3: describing incidents and intelligence

During an investigation, tagging observed behavior with technique IDs makes reports comparable and easier to share with vendors, CERTs and ISACs. It also makes it easier to search for related public reporting. Keep the tags at the most specific level supported by evidence: if you saw a full program download, tag T0843.001 rather than only the parent. The OT incident response guide covers which evidence sources support those conclusions.

Use case 4: exercises and purple teaming

Pick a published campaign, extract its techniques with the jq query above, and build a tabletop or a lab exercise around it. For a safety system scenario, use the Triton technique list; for a grid scenario, use the 2016 Ukraine campaign. Exercise the detections you claimed in your Navigator layer against a test bench, never against a live process.

Limits to keep in mind

  • ATT&CK describes observed behavior, so it lags novel techniques and reflects what has been publicly reported. Absence from the matrix does not mean a technique is infeasible.
  • Coverage is not risk. Covering many techniques with weak detections is worse than covering the few that matter for your consequences.
  • Impact needs engineering input. The Impact tactic only becomes meaningful when process engineers explain what Loss of Control means for a specific unit. The OT vs IT security guide explains why that consequence view drives OT priorities.

Use ATT&CK for ICS alongside your protocol knowledge (see ICS protocol security), your remote access design (secure remote access for OT vendors) and the program structure in the NIST SP 800-82 overview, and it becomes a practical planning tool rather than a poster on the wall.

References

Related articles

Build an OT incident response plan around safety and process continuity: roles, decision rights, playbooks, evidence from PLCs and HMIs, and verified recovery.
Build an OT asset inventory safely: SPAN and TAP placement, Zeek with CISA ICSNPP parsers, Wireshark filters for ICS protocols and which attributes to record.
Factual case studies of Stuxnet, the 2015 and 2016 Ukraine grid attacks and TRITON: how each attack worked, mapped to ATT&CK for ICS, and the defensive lessons.