AgentInterdict blog
Shorter, practical pieces on running AI agents safely. New posts most days. Each one targets a real question people hit when they put agents in production.
Latest posts
Tool Hijacking: When Your Agent's Tools Turn Against You
An attacker doesn't need to control your agent. They just need to control which tool it calls, and h
MCP Security: What You Actually Need to Know
Model Context Protocol is how agents talk to tools. It's also a new attack surface. Here's what MCP
Why Your Agent Needs Provenance Tracking
The scariest agent attacks aren't the agent going rogue. They're a poisoned document quietly steerin
Memory Poisoning vs. Prompt Injection: Know the Difference
They're often lumped together, but memory poisoning and prompt injection are different attacks with
What Is Prompt Injection, Really?
Prompt injection is the most misunderstood attack in AI security. Here's a plain-language breakdown
Why Agent Memory Is a Security Boundary
Most people treat agent memory as a data store. It's not. It's a security boundary, and the moment y
Why Prompt Filtering Alone Is Not Enough
Filters catch known attacks. Boundaries stop the ones you haven't seen yet. Here's the difference.
Fail-Closed vs. Fail-Open: The Decision That Defines Your Agent's Security
When your enforcement layer can't decide, what happens? That one choice determines whether you have a security boundary or a suggestion box.
Why a blog?
Most agent-security writing is either academic papers or vendor marketing. This blog is neither. It's the notes from actually running autonomous agents with memory and tools, and the enforcement layer that keeps them from being a liability. If you run agents, some of this will save you from a bad day.
New to the topic? Start with the core doctrine, then watch an attack get interdicted live.