- 15 min read
Agent guardrails: guard the actions, not the text
A chatbot that says something wrong is a bad day. An agent that does something wrong is a liability. Part 2 of our guardrails series moves from safe text to safe actions — the agent-specific threats (tool misuse, goal hijacking, excessive agency) and the controls that actually contain them: least privilege, approval gates, typed tools, and blast-radius limits.
AgentsGuardrailsSecurityTool UseArchitectureProduction - 16 min read
LLM guardrails are not one thing — they're a stack
Prompting is asking nicely. Guardrails are seatbelts, brakes, traffic lights, insurance, and the bouncer at the door. A practitioner's tour of the real guardrail stack — input, retrieval, output, and action — the frameworks that fill each layer, and where they bolt onto a portable agent architecture.
GuardrailsSecurityAgentsArchitectureProduction