agentic-ai
offense
Prompt Injection in Agentic AI Systems
Prompt injection has become the most dangerous attack vector against autonomous AI agents — and most teams building with LLMs still have no idea how to defend against it.
Articles
Insights, research, and guides from the Lost Edges Security team — covering security engineering, agentic AI, threat research, and offensive security.
offensive-security
red-team
Red Team Lessons: Bypassing EDR with Living-off-the-Land Techniques
During a recent internal red team engagement, our offensive security team achieved SYSTEM-level access on a hardened endpoint in under 4 hours — without dropping a single binary.
security-engineering
cloud
Threat Modeling Modern Cloud-Native Architectures
Microservices, service meshes, and ephemeral infrastructure have fundamentally changed the threat model for cloud applications. Here's how to adapt your threat modeling practice.
threat-research
apt
APT Group Analysis: Tactics of State-Sponsored Threat Actors
A technical deep-dive into the tooling, infrastructure, and behavioral patterns of three active nation-state threat groups targeting financial services and critical infrastructure.
security-engineering
zero-trust
Zero Trust Is Not a Product: Building a Mature Access Control Program
Every vendor sells 'zero trust.' Almost nobody actually practices it. Here's what a genuine zero-trust access control program looks like from the inside.