Azure penetration testing

Entra ID actor tokens, PRT replay, managed-identity over-privilege.

Entra ID actor-token impersonation (CVE-2025-55241), Primary Refresh Token replay via roadtx, device-code phishing through phantom device registration, managed-identity over-scope chained from IMDS SSRF, workload-identity federation issuer tampering, and Azure Monitor SSRF to token theft (CVE-2025-62207), tested by hand. Every finding ships with a working proof-of-exploit, code-level fix guidance, and a re-test.

See the Entra ID attack paths
Entra tenant graph, actor token, PRT replay, managed identity, KeyVault, subscription Owner pivots labelled.

Entra-first

Actor tokens, PRT replay, device-code phishing, tested against your tenant by hand.

Identity to Owner

Managed-identity over-scope and federation tampering chained to subscription Owner.

Re-test included

We verify your fixes at no extra cost. One engagement, closed loop.

Why now

Entra ID actor tokens cross tenants, your blast radius isn't your subscription.

Trusted by security teams across Fintech, SaaS & Education, Enterprise & Telecom, Security & Critical Infrastructure

Airbase
Quiltt
Pacvue
Imagine Learning

On record

  • CREST accredited
  • AICPA SOC 2 Type II
  • ISO/IEC 27001

What we test

Six surfaces. Six named bug classes.

These are not generic categories, they are the primitives our Azure pentesters chain into engagement findings.

Entra ID actor tokens
Undocumented service-to-service actor tokens accepted by legacy AAD Graph without source-tenant validation, cross-tenant Global Admin (CVE-2025-55241).
Primary Refresh Token replay
Extract CloudAP-protected PRTs from a joined host, replay via roadtx to mint MSGraph tokens that satisfy MFA and conditional access.
Device-code & CA bypass
Device-code phishing chained with phantom-device DRS registration marks the attacker workstation compliant, conditional-access policy waved through.
Managed identity over-privilege
Workload SSRF reaches IMDS, lifts a system-assigned identity with Contributor at subscription scope, then pivots tenant-wide (CVE-2025-62207).
Workload identity federation
Swap the federated-credential issuer URL on an Entra app, persistent service-principal access without a stored secret, no rotation signal.
Subscription & KeyVault RBAC
Owner or User Access Administrator at subscription scope plus permissive KeyVault access policies, reveals secrets, cert private keys, AKV-stored SAS tokens.

Beyond config

One phish becomes tenant Owner.

Most Azure security tooling stops at the config-audit layer, checks IAM bindings, surfaces misconfigurations, emits a CSV. That's where a real pentest begins. A phished employee gives up a Primary Refresh Token; using roadtx and CloudAP, the pentester mints a fresh access token that satisfies MFA and conditional access because a phantom DRS registration makes the device look compliant. From an internal workload SSRF (the same primitive that produced CVE-2025-62207 in Azure Monitor), the team reaches IMDS and lifts a system-assigned Managed Identity scoped Contributor at the subscription. That identity rotates a KeyVault access policy, swaps the issuer on a privileged Entra app's Workload Identity Federation credential, and the chain closes at subscription Owner, the same class of trust abuse behind CVE-2025-55241. One starting credential, four primitives, tenant-wide compromise. Config audits don't catch chains; pentests do.

See BugDazz, SecureLayer7's autonomous pentest →
Four-step chain diagram, phished PRT, conditional-access bypass via phantom DRS device, Managed Identity pivot through IMDS SSRF, and Workload Identity Federation issuer swap closing at subscription Owner.
Four-step chain diagram, phished PRT, conditional-access bypass via phantom DRS device, Managed Identity pivot through IMDS SSRF, and Workload Identity Federation issuer swap closing at subscription Owner.

Meet your Azure lead

Hands on every Azure engagement.

John Dill

vCISO at SecureLayer7

15+

Years in offensive security

150+

Engagements led to date

99.99%

On-time engagement delivery

John scopes Azure engagements against your Entra tenant, subscription topology, and hybrid-identity boundary. He sits in the room from kick-off through findings review and re-test.

  • Scopes Entra ID, conditional access, and managed-identity paths against your real risk model.
  • Walks every AKS, Key Vault, and workload-identity finding live with your team.
  • Drives remediation review and re-test until every tenant-wide path is closed.
SL7 Lab. Published CVE research.
John Dill, vCISO at SecureLayer7

Ready to scope an Azure pentest? Book 30 minutes with John to walk through your Entra tenant, subscription layout, and timeline.

Book a 30-min call

AZURE PENTEST ENGAGEMENT.

How we run an Azure pentest. Tenant-wide, closed-loop.

Threat-modelled to your tenant, Entra graph, and subscription topology. Aligned with Microsoft's pentest rules-of-engagement, not a template we run against every cloud.

  1. 01
    Scope & rules-of-engagement
  2. 02
    Read-only access provisioning
  3. 03
    Tenant reconnaissance
  4. 04
    RBAC review
  5. 05
    Active exploitation
  6. 06
    Post-exploitation & blast-radius
  7. 07
    Reporting & remediation
  8. 08
    Re-test & closure

Procurement questions

What buyers ask before signing.

Seven questions procurement teams send before signing an Azure pentest SOW. Answered against Microsoft's rules-of-engagement and your auditor.

Show all 7 questions

Have a procurement question not listed here?

Tested by industry.

The bug classes named below come from real engagements in each sector. Pick the closest fit.

Tech SaaS

Multi-tenant SaaS on Azure, Entra ID drift, conditional-access bypasses.

FinTech

Banking workloads on Azure, Key Vault boundaries, M365 + Defender attack paths.

HealthTech

HIPAA-aligned Azure tenants, PHI in Storage, Healthcare APIs on Azure.

Sample Azure pentest report, kill-chain · evidence · remediation

Sample engagement report

See what arrives in your inbox.

A pre-vetted Azure engagement sample: full vulnerability narrative, working proof-of-exploit traces, and Bicep, ARM, or Terraform fix guidance you can hand to your platform team. Sent on request after a 5-minute scoping call.