Services Authorised. Focused. Evidence-led.
Authorised penetration testing across applications, cloud, internal and AI systems. We document findings for engineering and security teams to investigate and act on.
From a focused application test to a connected, multi-layer assessment.
01 / Identity & access
A stolen session. A newly disclosed flaw. An exposed workload credential. The opening may differ, but the next question is the same: what can it reach? We test the permissions that should keep one foothold from becoming something larger.
A trusted identity should not be a master key.
Application permissions
A valid login should not become another customer's data. We test where your application trusts a user, object or request more than it should.
The evidenceReproduction steps tied to affected endpoints, roles and business impact.
Cloud identity
A workload can be secure in isolation and overprivileged in context. We trace identity and policy relationships across AWS, Google Cloud and Azure.
The evidenceDemonstrated privilege paths and the policy or configuration changes that address them.
The path shows where to tighten access, limit privilege and identify activity your defenders should be able to see.
02 / Delegated authority
AI & LLM red-teaming
A hostile instruction can arrive in a document, a search result or a tool response. The security question is what happens when that content meets an agent with real permissions.
Reproducible test cases connecting the input, permissions, tool behaviour and observed impact. Your team can see which boundary failed and what needs to change.
03 / Containment
Access to one part of the environment should not mean access to all of it.
Internal & assumed-breach testing
We begin with an agreed level of access and test how far it can travel. Identity controls, network segmentation and credential handling are tested together, not as separate checklists.
Alongside the demonstrated path, we identify relevant telemetry and detection improvements. The work should explain both how access was gained and where defenders can intervene.
A boundary matters when it limits what comes next.
04 / Verified repair
Remediation & retesting
A patch is a change, not yet a result. We revisit the demonstrated issue and agreed bypass cases. Then separate what is resolved, what remains open and what could not be tested.
The report follows the agreed scope. Retesting is recorded when included in the engagement.
Assets, permissions, test window and limitations.
Prerequisites, reproduction steps and observed impact.
Recommended changes and useful defender signals.
What was retested, the result and what remains.
05 / Agree the boundaries
We test with your permission, not at the expense of your operations. Scope and safety are part of the engagement, not small print.
Named assets, agreed techniques and third-party exclusions before testing starts.
Test windows, stop conditions and an escalation contact agreed with your team.
Agree how sensitive evidence is collected, shared, retained and removed.
Unexpected access, a third-party boundary or a risk to availability triggers a stop and review with your designated contact.
How we work safelyStart with a conversation
Tell us about the system, the decision you need to make and your timing. We will work with you to define a useful scope.
Share a high-level outline first. An NDA can be arranged before sensitive details.