Loading...
Services
We build the infrastructure that makes AI-assisted security work defensible.
Security teams are adopting AI IDEs. The risk is not that they use them — it is that there is no consistent methodology behind them, no record of what the AI was asked or what it produced, and no way to prove the instructions came from a trusted source.
That gap is a liability in any regulated environment where internal audit, regulators, or clients ask how AI was used in the assessment. When they ask, “files on SharePoint” is not a sufficient answer.
We design and deliver the engineering layer that closes those gaps: authenticated playbook delivery, cryptographic audit trails, supply-chain provenance controls, and the threat modelling, architecture, and governance work that makes AI-assisted security defensible — not just functional.
Engineering Infrastructure
Consistent, verified playbook delivery to any AI IDE your team uses.
What it is
A containerised Model Context Protocol (MCP) server that delivers peer-reviewed security instructions, workflows, and templates directly into Claude, Cursor, and GitHub Copilot. Every consultant uses the same content, from the same verified source, every time.
A tamper-evident, signed record of every AI-assisted assessment action.
What it is
An append-only, hash-linked audit log written with every invocation of the Playbook MCP Server. Each entry is signed using a key whose private material never leaves the key management service. Any auditor can verify the chain without holding signing capability.
Assessment and Testing
Structured threat modelling for systems where AI is part of the attack surface.
What it is
A structured threat modelling and risk assessment service for AI-integrated systems — covering the threat vectors that traditional frameworks do not fully address, including prompt injection, training data poisoning, model extraction, and inference-time manipulation.
Adversarial testing for systems that standard penetration testing does not cover.
What it is
Structured adversarial testing of AI-integrated security systems — the attack paths that automated decision systems introduce, sitting outside conventional penetration testing scope.
Design, Build and Governance
Architecture that is fit for purpose from the start, not retrofitted after the audit.
What it is
Architecture advisory and design services for organisations building or adopting AI-assisted security tooling — from trust boundary definition through to full target-state architecture, aligned to TOGAF views and your governance standards.
From design to working tooling — hands-on engineering for AI-integrated security systems.
What it is
Hands-on implementation for organisations moving from architecture to production AI-assisted security tooling. Our engineers have built production AI-integrated security systems and can work alongside internal teams or deliver end-to-end.
The governance framework your AI security tooling needs before the regulator asks for it.
What it is
A governance and policy development service for organisations deploying AI into security functions. We build the policy framework, evidence standards, and incident procedures needed to operate AI-assisted security tooling under regulatory scrutiny — before the audit, not in response to it.
We design the controls before the audit finds the gaps.
What it is
An advisory service for organisations building or deploying AI-assisted security tooling. We cover architecture, supply-chain trust, identity and access design, audit trail requirements, and the threat model questions that vendor documentation does not answer.
Origin
This capability was not developed for a service catalogue. We built it to solve a problem we had in our own assessment practice: no defensible record of what AI was asked, what it produced, or whether the instructions came from an authorised source.
We deployed it internally first. We then deployed it with clients. We are describing it publicly now because clients asked us to.
Deployments
Currently deployed across two client organisations in regulated environments and in active use within the CyberTeam practice. In all three deployments, teams observed a measurable reduction in assessment time and a consistent improvement in output quality and audit trail completeness.
We are not publishing specific metrics at this stage. If you want numbers, ask us on a scoping call — we will tell you what we have seen and what is attributable to the tooling versus other factors.
Client
“We needed to demonstrate to our internal audit team that AI-assisted assessments followed a consistent, traceable methodology. CyberTeam built the infrastructure that made that conversation possible.”
We built this tooling to solve a problem in our own assessment practice before we offered it to anyone else. CyberTeam runs on the Playbook MCP Server and Attestation Chain. When we tell you how the controls behave under audit pressure, it is because we have been in that position — not because we modelled it.
Our work spans cryptographic audit trail design, threat modelling, adversarial testing, hands-on build, and governance policy. We do not hand off between specialisms or subcontract the implementation. The same team that designs the attestation chain can implement it and walk your audit team through the verification process.
Recommendations default to APRA CPS 234 and Essential Eight — the frameworks that regulated organisations in this market are actually measured against. Where your obligations differ, we map to your confirmed control catalogue. Generic best-practice documents are not what we deliver.
We start with a 45-minute scoping call: your AI tooling landscape, current controls, and where the audit or regulatory exposure sits. You will leave with a clear view of your priority risk areas and a proposed engagement scope. There is no obligation beyond the call.
If you are not ready for a call, request the technical overview — a single document covering the Playbook MCP Server and Attestation Chain architecture, controls design, and deployment model.