You call, you enforce.
Rather govern from inside your own code than route through a proxy? Call the SDK before an agent acts — it returns allow, hold or block, and your code enforces the answer. On PyPI.
For teams with their own stack.
Use the Gateway (a proxy) when you want governance live in minutes with no code. Use the Verdict SDK when you’d rather keep enforcement in your own code — three reasons teams do:
You already have a gateway
Kong, an API layer, a platform you have matured for years — keep it. Call the Verdict and enforce the answer where your traffic already flows.
Governance inside your product
Ship it in your own code without handing traffic to a proxy. The decision comes back in milliseconds; you act on it, in your own control plane.
The same rulebook as the proxy
The Verdict runs the exact rules the Gateway does and writes the same record — so you can mix the SDK and the proxy, team by team.
Keep reading
Runtime Is Where Agent Security Gets Real
A valid credential doesn't make a live action valid.
Give Your Agents Less. On Purpose.
Everyone races to give AI agents more power.
Prompt Injection Is Not an AI Problem. It's an Agent Problem.
Most content about prompt injection focuses on making the model say bad things.