- Market
- Saudi Arabia
- Written for
- CISO, CTO
- Decision
- Find the reachable attack paths in an AWS or Azure environment
- Languages
- EN · AR
Cybersecurity credentials
NCA registration
Registered cybersecurity service provider

ISO/IEC 27001
ISO/IEC 27001 certified
The escalation
How an attack walks from a leaked key to your production data
Each step is legitimate on its own — an identity using permissions it was granted. The engagement is showing that the granted permissions, followed to their conclusion, add up to a path nobody intended.
1 · A leaked or over-permissioned credential
A key committed to a repository, sitting on an instance, or held by a developer role that was granted broad access "for now" and never narrowed. The starting identity, and what it is already allowed to do.
2 · Enumerate what that identity can reach
Reading the permission policies, role trust relationships, storage, secrets and services the identity can see — the picture the console never shows on a single screen. A map of reachable permissions across the tenant.
3 · Escalate through misconfiguration
Assume-role chains, a pass-role permission handed to a service, or a policy that lets the identity attach a stronger one to itself — privilege gained without a single exploit. A higher-privileged identity reached by design, not by breaking anything.
4 · Reach the data or the control plane
Read the storage bucket, query the database, or take administrative control of the account — stopping at the point that demonstrates impact, inside the boundary agreed in writing. Demonstrated impact, with the exact chain and the stop point stated.
Test versus review
A configuration report tells you what is set, not what can be reached
A cloud configuration review reads your settings against a benchmark and reports where they diverge — a public bucket, an over-broad policy, a missing log. It is genuinely useful and it should be run continuously. What it cannot tell you is which of a thousand findings actually chains into a path to your data, because it evaluates each setting in isolation and an attack is the combination of several of them followed in sequence.
A penetration test starts from an identity and follows the permissions until they either dead-end or reach something that matters, exactly as an attacker would. The two are complements, not substitutes: the review keeps the estate clean day to day, and the test proves, on a specific date, whether the settings that survived the review still add up to a reachable path. Buying only the review leaves you with a list you cannot prioritise; buying only the test leaves the estate to drift between engagements.
What surfaces
The cloud findings that a checklist review never surfaces
Assume-role chains
An identity permitted to assume a role that is permitted to assume another, ending somewhere it was never meant to reach. Each grant is defensible alone; the chain is the finding.
Pass-role and self-escalation
A permission to hand a role to a compute service, or to attach a policy to an identity, that becomes a route to administrator — a privilege-escalation primitive a benchmark scan reads as an ordinary setting.
Trust extended to an outside principal
A resource shared to another account, or a role that trusts an external identity, configured deliberately and forgotten — reachable from outside the tenant, and read by a config scan as intended.
Secrets in reachable places
Keys in an instance profile, a build log, an environment variable or a shared parameter store — credentials that a static review of the configuration never opens, and that shorten the whole path.
Our role
What we exploit in your tenant and what your platform team fixes
- 01
We start where an attacker would
From a low-privileged identity or a leaked-style credential you supply, not from administrator access, because a test handed admin rights only proves it can use them.
- 02
We give the chain, not the catalogue
The report is the ordered path from starting identity to impact, so remediation can break the chain at its cheapest link rather than triaging hundreds of unranked misconfigurations.
- 03
Your platform team owns the estate
Because the cloud changes on the next deployment, we leave the finding patterns and the escalation primitives so your team can catch them in review going forward, not only in the next test.
Which accounts and services are in scope, and where the boundary sits between your resources and a provider's managed platform, are yours to set; we walk the identity paths inside that boundary and hand back a chain your team can close and then keep closed as the environment changes.
Common questions
What buyers ask
Is cloud penetration testing the same as a cloud configuration review?+
No, and they answer different questions. A configuration review evaluates each setting against a benchmark and produces a list of divergences, which is ideal as a continuous control. A penetration test starts from an identity and follows the permissions until they reach something that matters, proving which of those settings actually chain into a path to your data. The review keeps the estate clean between tests; the test tells you, on a date, whether what survived the review still adds up to a way in. Most organisations need both, and buying only the review is the more common mistake.
Do we need our cloud provider's permission to test?+
For your own resources, the major providers now permit customer-authorised testing without a request in most cases, but the boundary matters: their managed services, another tenant, and anything shared with a partner are outside what your authorisation covers, and denial-of-service style techniques are prohibited regardless. We confirm the specific provider policy for the services actually in scope during scoping, because a permission that covers your own compute rarely extends to a managed database or an identity service, and discovering that mid-test wastes days.
Can you test a production AWS or Azure environment safely?+
Yes, because most of the work — enumerating permissions, mapping trust relationships, tracing escalation paths — is read-only and carries no operational risk. The risk sits in a small set of active steps, such as actually assuming a role or writing to a resource to prove control, and those are agreed in advance, bounded, and stopped at the point impact is demonstrated rather than carried further. Where a step could disrupt a live service, it is scheduled into a window or replaced with a read-only proof, and the report states which was done.
In this cluster
Next step
Send your cloud accounts and roles
A scoped cloud test with a starting identity, an agreed boundary in your account, and a report that gives the exact escalation chain rather than a list of misconfigurations to triage.
