Field Notes.
Technical briefings from the Basalt red team and defense division. AI security, penetration testing, zero trust, incident response and compliance for New Zealand and Australian organisations.
Red Teaming RAG Systems: Data Exfiltration and Context Poisoning
Retrieval augmented generation widens the attack surface. How we red team RAG pipelines for data exfiltration, context poisoning and cross-tenant leakage.
MDR vs Managed SOC vs SIEM: A Buyer Guide Without the Black Box
MDR, managed SOC and SIEM compared honestly. How to choose detection and response that does not black-box your detections or lock you into a vendor.
Threat Intelligence That Drives Detections, Not PDF Reports
Most CTI ends up as a report nobody reads. How to build a threat intelligence pipeline that maps adversary behaviour to detections in your SIEM and XDR.
Post-Quantum Cryptography: Building a Migration Plan Before the Deadline
Harvest now, decrypt later is already happening. A practical PQC migration plan covering cryptographic inventory, crypto-agility and the NIST FIPS 203, 204, 205 standards.
ISO 27001:2022 Certification: A Readiness Guide That Skips the Theatre
ISO 27001:2022 without the parallel paperwork. How to map the Annex A controls to systems you already run and reach certification in months, not years.
NZISM and the Privacy Act: A Cyber Security Compliance Guide for NZ
A clear guide to the NZISM and the Privacy Act 2020 for New Zealand organisations, mapping the obligations to security controls you can actually operate.
The Essential Eight Maturity Model: A Pragmatic Path to Level Two
The ACSC Essential Eight, decoded. A practical guide to the eight mitigation strategies and how to reach Maturity Level Two without stalling your business.
Inside a Source Code Security Audit: SAST, SCA and Manual Review
What a thorough code security audit covers beyond automated scanning: secrets, authentication, injection, deserialisation and supply chain risk, with fix guidance.
Secure SDLC in Practice: Shift Left Without Slowing Down Engineering
A secure SDLC that engineering teams adopt rather than route around. Threat modelling, paved roads, CI security gates and AppSec champions that ship.
ITDR Explained: Detecting Identity Attacks Your SIEM Misses
Identity is the modern attack surface. How ITDR detects account takeover, session hijacking, MFA fatigue and lateral movement that traditional SIEM rules miss.
Cloud Security Posture: Cutting Through CSPM and CNAPP Noise
CSPM and CNAPP tools generate thousands of findings nobody owns. How to tune cloud security posture management to attack paths that actually matter.
Building an Incident Response Playbook Your Team Will Actually Use
A reusable incident response playbook structure covering roles, severity, containment, evidence handling and communications, aligned to NIST 800-61.
Ransomware Readiness: The Controls That Decide Whether You Recover
Ransomware readiness is not one backup. A practical assessment of the identity, segmentation, backup and response controls that determine recovery time and impact.
Zero Trust Architecture: An Implementation Roadmap That Ships
Most zero trust projects stall on the whiteboard. A phased roadmap mapped to NIST 800-207 that sequences identity, device posture and segmentation around real systems.
What a Real Web Application Penetration Test Looks Like
Beyond the automated scan: a transparent look at our web app penetration testing methodology, from recon and auth testing to exploit chaining and retest.
Model Context Protocol (MCP) Server Security: A Hardening Checklist
MCP servers are a new and exposed attack surface for AI agents. A concrete hardening checklist covering auth, tool poisoning, scope creep and supply chain risk.
Securing AI Agents: Tool Abuse, Confused Deputies and Blast Radius
When an LLM can call tools, the threat model changes. How to scope agent permissions, contain tool abuse and stop confused-deputy attacks on agentic systems.
A Working Taxonomy of LLM Jailbreak Techniques
From role-play and token smuggling to crescendo and many-shot jailbreaks, a categorised reference of how guardrails fail and how to test for each class.
Indirect Prompt Injection: How Untrusted Content Hijacks Your LLM
Indirect prompt injection turns documents, web pages and emails into attacker control surfaces. Here are the defense patterns that actually hold in production.
The OWASP LLM Top 10 (2025), Explained With Real Mitigations
A practical walkthrough of the 2025 OWASP Top 10 for LLM applications, with concrete controls for each risk class from prompt injection to unbounded consumption.