Security testing services

What we test

• Authentication, session lifecycle, and token handling
• Authorisation across every user role and tenant boundary
• Business-logic abuse and workflow bypass
• Injection, input validation, and output encoding
• API endpoint enumeration and object-level authorisation
• File upload handling and server-side request forgery
• Transport security and security header configuration

What you receive

• Executive summary written for non-technical stakeholders
• Per-finding reproduction steps a developer can follow
• CVSS v3.1 vectors with impact and likelihood rationale
• Specific, actionable remediation guidance
• A verification retest and updated remediation-status report

Typical engagement

Two weeks of testing for a mid-sized application, followed by a remediation window and retest. Exact duration is confirmed during scoping.

What we test

• Direct and indirect prompt injection
• System-prompt and context-window disclosure
• Insecure output handling leading to XSS or code execution
• Excessive agency in tool-calling and agent workflows
• Sensitive data leakage through model responses
• RAG pipeline poisoning and document-boundary bypass
• Rate limiting, cost abuse, and denial-of-wallet exposure

What you receive

• Reproducible attack payloads with observed model behaviour
• Severity assessment reflecting real business impact
• Guardrail and architectural recommendations, not just filters
• A verification retest after mitigations are deployed

Who this is for

Teams shipping customer-facing chat, retrieval-augmented search, or agentic features — particularly where the model can access internal data or trigger actions.

What we review

• IAM roles, policies, and privilege-escalation paths
• Object storage exposure and access-control configuration
• Network segmentation, security groups, and public surface
• Secret storage, rotation, and hardcoded credential exposure
• Logging, monitoring, and audit-trail coverage
• Model endpoints, inference APIs, and vector database access
• Encryption at rest and in transit

What you receive

• Prioritised findings with exploitability context
• Configuration-level remediation steps per provider
• Benchmark alignment summary
• A verification retest after changes are applied

Providers covered

Amazon Web Services, Google Cloud Platform, and Microsoft Azure — with current security certifications held across all three.

What it covers

• External attack-surface discovery and enumeration
• Known-vulnerability and outdated-component identification
• Service, version, and configuration exposure
• Certificate and transport security posture
• Manual triage of every reported result

How it differs from a pentest

A vulnerability assessment tells you where known weaknesses exist across a wide surface. A penetration test proves what an attacker can actually chain together and achieve on a specific target. Most organisations benefit from both — assessment for coverage, testing for depth.

Why manual triage matters

Automated tools produce substantial false positives. Every result we report has been manually verified, so your engineering time goes to real issues.