Free resource · 16 controls · MIT / CC BY 4.0

The AI Agent Security Checklist

A practical control checklist for teams shipping LLM agents, the kind with tool access, credentials, and the ability to take real actions on real systems. Every control states what to check, why it matters, how to test it, and what the hardened version looks like in real configuration or code. Mapped to OWASP’s Top 10 for Agentic Applications and Top 10 for LLM Applications, MITRE ATLAS, and NIST’s AI Risk Management Framework. That makes findings here citable in an enterprise security questionnaire, not just useful in a design review.

The checklist groups AI agent controls into input handling, identity, tools, data, and observability.
Start with the control domains that define what an agent may read, call, and change.

How to use this

Shipping an agent in the next month? Work the nine Baseline controls below first. They cover the mechanism that turns a successful prompt injection into an incident rather than an annoyance: a scoped credential, a validated tool argument, and a human approval gate on anything that writes.

Answering a security questionnaire? Each control’s framework mapping gives you a citable identifier: an OWASP ASI or LLM risk, a MITRE ATLAS technique, or a NIST AI RMF subcategory, rather than a vague claim of "we thought about AI security."

Reviewing someone else’s system? Every control’s “How to test” section is written to be executable by a reviewer who did not build the thing and cannot ask the person who did.

Baseline: start here

The controls that turn a successful attack into a non-event instead of an incident.

TOOL-01Tool authorizationAgent credentials are scoped to the tools the agent actually uses
TOOL-02Tool authorizationTool arguments are validated in code, and the model never supplies the target
IDN-01Identity & blast radiusThe agent does not share a credential with a human or an application
INP-01Prompt injectionAll content from outside the agent’s workspace is treated as adversarial
INP-02Prompt injectionA trust boundary separates ingesting external content from taking privileged action
HIL-01Human-in-the-loopProduction writes and infrastructure changes are gated on human approval
OBS-01ObservabilityTool calls, arguments, and results are logged well enough to reconstruct an incident
RES-01Resilience & abuseLoop, spend, and rate limits bound the blast radius of a misbehaving agent
OUT-01Output handlingModel output is never rendered or executed downstream without contextual encoding

Want someone else to run this against your system?

This checklist is what a self-review looks like. An AI Security Review is what it looks like when I run it against your actual agent, with findings ranked by exploitability and hardened-by-example fixes, in three weeks.