POLICY ENGINE

Build a policy that adapts to your agent.

Every business runs agents differently. Kernel doesn't ship one-size-fits-all rules. Configure it to your scale, your jurisdictions, and your tolerance for autonomy.

STEP 1 · DESCRIBE THE TASK

Tell Kernel what your Payment Execution Agent does.

Write the agent description in the same words you'd give a new hire on day one. The clearer you are about scope, scale, and reversibility, the sharper the risks and conditions Kernel proposes.

517 chars · Kernel extracts intent, blast radius, and reversibility signals from this description to seed step 2.

THREE WAYS TO BUILD A POLICY

Pick the path that fits your team. The engine above is the default.

A

Pre-built templates

Your agent type probably already exists. Select it above, configure your thresholds, and deploy. No policy writing required.

Fastest path
B

Use our engine

Walk through the four steps above. Configurable conditions, production-ready YAML at the end.

C

Work with us

For complex or heavily regulated deployments, the Kernel team reviews and co-develops your policy before it goes live.

POLICY SCHEMA V1.0 · KERNEL AI LABS S.L. ·