Security

Every Security article we've published — field notes on building and shipping production AI systems, written by the engineers who build them.

4 articles

Moses Otieno13 min read

Break the lethal trifecta: six agent designs that survive prompt injection

You can't reliably detect prompt injection, so design agents where it can't trigger anything that matters. How to find the agents at risk, the six patterns that fix them, and how to choose between them.

Sep 15, 2026 Read article

Moses Otieno12 min read

The data your agent reads can give it orders

Indirect prompt injection is live on the open web. You can't filter your way out of it - but you can make a successfully injected agent unable to do anything that matters on its own.

Sep 12, 2026 Read article

Moses Otieno11 min read

Concurrency-safe isn't tenant-safe - the shared state hiding in your agent

A mutex makes a package-level handoff slot race-free without making it safe. Why agent servers reintroduce an old mistake, and how keying every handoff to its turn closes it for good.

Sep 11, 2026 Read article

Moses Otieno12 min read

Letting an agent act as you - safely

Autonomy without guardrails is a liability. The three decisions that make agent actions shippable: forwarded identity, human-in-the-loop on writes, and an immutable audit trail - at scale.

Aug 5, 2026 Read article