Guardrails

Constraints put around an agent's behavior to prevent harmful, out-of-scope, or unintended actions. Can be implemented as input/output filters, tool restrictions, or a separate validator model.

📌 Example: A guardrail blocks any tool call that would delete a database record without a confirmation step, regardless of what the agent decides.

🛠 In practice: Guardrails are the safety net between autonomy and accountability. The more autonomous the agent, the more important robust guardrails become.

• motivates Prompt InjectionThe risk of prompt injection is one of the primary reasons guardrails are essential in agentic systems.
• is a form of Human-in-the-LoopRequiring human approval at key steps is a behavioral guardrail, implemented at the workflow level rather than as a filter.