What is AI agent security?
AI agent security protects systems where AI agents can call tools, access data, trigger workflows, or execute actions. It focuses on preventing unsafe behavior before the action happens.
What is runtime enforcement?
Runtime enforcement means checking an action while the system is running, before the agent executes it. This is different from reviewing logs after the fact.
What is prompt injection protection?
Prompt injection protection detects and blocks malicious instructions that try to override the agent's original rules, leak data, or force unsafe tool use.
What are signed audit receipts?
Signed audit receipts are cryptographic records of security decisions. They show what action was checked, what decision was made, and whether the receipt has been modified.