Critical infrastructure

Test how far a corporate foothold carries an attacker toward control.

Your external ranges to the last IT system before the control zone.

No PLC, HMI or SCADA is touched — this run stays IT-side. What comes back is the path, the credentials that opened it, and the line where the run stopped.

Book a demo
See the platform
Definition

What is it/ot boundary validation?

IT/OT boundary validation proves by execution how far an attacker who already holds a corporate foothold can move toward the industrial control network. It walks the credential and network paths between the zones on the live estate and records where each route stopped.

Reachability, not the segmentation diagram

Walks the credential and network paths between the corporate zone and the systems bordering control, and reports where each route was stopped short of the control network.

No PLC, HMI or SCADA is touched

Targets are authorised before the run starts, approval on every command is the default, and the run stops at the last IT-side host.

Emulate the real actors, mapped to MITRE ATT&CK

Runs the tradecraft behind real infrastructure intrusions, with technique IDs recorded on the operation graph as the run goes.

Every command is on the record

The operation ledger holds each command, its result and the artifacts it produced: an audit trail of the whole path and a stated stop line.

Volt Typhoon (PRC) — living off the land with native admin tools, pre-positioned inside IT to reach OT in energy, water and communications.

Volt Typhoon reaches OT through the IT estate. So does the run.

The operator logs in with your own credentials, works with your own admin tools, and waits at the boundary. Five stages, and the one where an authorised run stops.

  1. 1Initial access
    Attacker

    Logs in with valid accounts. No exploit, no malware.

    With Cracken

    Enumerates what actually answers on your authorised external ranges, and tests which of those entry points accept credentials.

  2. 2Living off the land
    Attacker

    Uses the admin tooling already on the host, so there is nothing to signature.

    With Cracken

    Runs the same native-tool tradecraft from an installed Tentacle and logs every command, so your team can match it against what detection saw.

  3. 3Credential access
    Attacker

    Dumps the domain database and reuses what it finds.

    With Cracken

    Reproduces the domain attacks against the live directory — roasting, ADCS abuse, delegation, coercion and relay — and records only the credentials it actually obtained.

  4. 4Lateral movement
    Attacker

    Moves toward the systems that border OT.

    With Cracken

    Attempts the crossing with the credentials the run already proved, and records where each route was stopped short of the control network.

  5. 5Pre-positioning
    Attacker

    Sits on the boundary and waits for the moment to disrupt.

    With Cracken

    Stops at the last IT-side host and hands over the ledger. Nothing touches a PLC, an HMI or SCADA.

One run

From the public edge to the last IT host before control.

The external domains and ranges in scope, one corporate segment with a Tentacle on a host inside it, and the Windows domain behind them. Nothing on the control side is a target. The run stops at the last IT-side host.

Written from the execution plan of the playbooks that run today — Domain Recon, Web App Pentest, Network Pentest, Cloud Pentest, AD Pentest. A run covers the target you authorise, under the policy you set.

  1. 01
    What ran

    Enumerate the authorised domains and external ranges: subdomains, certificate history, resolving hosts, then the open ports, services and versions on each public IP.

    Domain Recon playbook — ships
    What it established

    Which addresses in scope actually answer from the internet, and what is behind each one — as opposed to what the asset inventory lists.

  2. 02
    What ran

    Install a Tentacle on a host in the corporate segment. Discover hosts, enumerate services and versions, test what answers, and read the firewall rules and ACLs meant to contain them.

    Network Pentest playbook — ships
    What it established

    What a first foothold reaches inside the network, and which containment rules are enforced rather than documented.

  3. 03
    What ran

    Work the domain from that foothold: Kerberos roasting, ADCS abuse, delegation, ACL and GPO abuse, coercion and relay, cross-forest trusts. Every credential is reproduced against the live domain before it is recorded.

    Active Directory Pentest playbook — ships
    What it established

    Which identities a corporate foothold can obtain, proven by reproduction rather than inferred from a path map.

  4. 04
    What ran

    From the segments those credentials open, attempt the crossing into the next segment toward the control network, and record where the attempt was stopped.

    Network Pentest playbook — ships
    What it established

    Whether the segmentation between corporate IT and the systems bordering control holds against credentials the run already proved.

  5. 05
    What ran

    Stop at the last IT-side host. Hand over the operation ledger: every command, every result, every artifact, with ATT&CK technique IDs recorded on the graph.

    Operation ledger — ships
    What it established

    An audit trail of the whole path and a stated stop line.

What this did not prove: The run proves nothing about the control network itself. No PLC, HMI or SCADA is contacted, and no OT-side control is tested. Where it stops is a scope decision, not a control that held — whether that last host can actually reach a controller is left open.

Run the operator's tradecraft yourself, first.

A state operator has already run this against a utility like yours. Ask a general-purpose model for the same living-off-the-land tradecraft and it declines; Cracken's model writes it under authorisation, and the run still stops at the last IT-side host.

Move on credentials, not implants

Living-off-the-land tradecraft with the admin tools already on the host, so there is no malware for a signature to catch.

Cross the estate

Builds the credential path from a first foothold to the segments that border control.

?

Stop at the line

Writes the tooling for the crossing attempt against your live estate, then stops at the last IT-side host and records where it was stopped.

A state operator has already run this against a utility like yours. Run it yourself first, on your terms.
Questions

What a utility security team asks before authorising a run

Does Cracken touch our PLCs, HMIs or SCADA?

No. This run is scoped to the IT side: Domain Recon, Web App Pentest, Network Pentest, Cloud Pentest and Active Directory Pentest. An authorised run stops at the last IT host before the control network and records where it stopped.

We already inventory our external attack surface. What does an authorised run add?

It tests what answers, not what is listed. CISA told the water sector the same thing on 30 July 2026: "Even water organizations with mature cybersecurity processes should validate their external connections, as this targeting activity includes cellular modems installed by operators, vendors, or system integrators that may not be documented or included in routine attack surface scans." A run enumerates the authorised ranges and reports the service and version behind every port that responds. — CISA Urges Water and Wastewater Systems Sector to Protect OT Against Activity Targeting PLCs (Cybersecurity and Infrastructure Security Agency (CISA), 2026-07-30)

Seven federal agencies warned about Iranian-affiliated actors disrupting PLCs. Does this address that threat?

Partly, and only on the IT side. Advisory AA26-097A, issued by the FBI, CISA, NSA, EPA, DOE, CNMF and Treasury and last revised 22 July 2026, describes actors disrupting internet-connected OT devices across Government Services and Facilities, Water and Wastewater Systems and Energy, "resulting in operational disruption and financial loss." An authorised Cracken run establishes which addresses in your external ranges answer and how far a corporate foothold travels toward the boundary. It does not interact with a PLC. — AA26-097A: Iranian-Affiliated Cyber Actors Exploit Programmable Logic Controllers Across US Critical Infrastructure (Originally published 2026-04-07, last revised 2026-07-22, FBI, CISA, NSA, EPA, DOE, CNMF and Treasury, 2026-07-22)

Has Cracken run this against a utility already?

No. The published evidence is research, not customer results: the team's paper "Red-Teaming the Agentic Red-Team" (arXiv:2606.24496, 23 June 2026) reports a security analysis spanning twelve agentic tools, with code execution achieved in 97.8% of runs aggregated across ten agentic red-teams, and Project BlackSea, an open-source honeypot published under Apache-2.0. — Red-Teaming the Agentic Red-Team (Pasquini, Bazyli, Fedynyshyn, Sorokin) (arXiv:2606.24496, arXiv, 2026-06-23)

Does the run need a person on site, and can we stop it mid-run?

Nobody needs to be on site once the Tentacle is installed on a host in the target segment, and the same scope can be re-run after a firewall change from wherever your team sits. Targets are authorised before the run starts and nothing acts outside that list. Approval on every command is the default until you raise the ceiling, and the run can be stopped at any point.

Can it run inside our own environment, and does our data train a model?

Cracken deploys self-hosted, which is what regulated operators usually require. For self-hosted and sales-managed workspaces, model training is disabled, so workspace content is not used for model improvement. That statement is scoped to those two workspace types and does not extend to every plan — check the scope against the plan you buy.

If you run the operation

See the engine underneath this.

Tentacles, the Cybergraph, the approval gate every action passes through, and the operation ledger that records what ran. Critical infrastructure is one playbook on top of that engine.

If you own the risk

Start from the exposure, not the technique.

One playbook answers one question. The case for validating exposure at all — why a scanner score is not a finding, and what changes when something proves the path instead of ranking it — is the argument this page assumes.

Use cases

Playbooks for Critical infrastructure

See all use cases

Find the IT routes that lead toward your control network.

Scope a run that stops at the IT side of the boundary.

Book a demo