Welcome to the agent::offense research log.
This is where we publish write-ups, tooling releases and offensive-security research — with a sharp focus on the newest attack surface: autonomous AI agents.
What you will find here
We break systems for a living: web and API applications, cloud and infrastructure, full red-team engagements, and increasingly the LLM-driven agents that now act with real permissions inside those environments.
Coming up
- Prompt-injection techniques from live engagements
- MCP and tool-chain security deep-dives
- RAG pipeline poisoning and data-exfiltration paths
- Classic pentest tradecraft that still pays off
Stay tuned. ./subscribe