Agent Rules

Prevent agent production incidents with opinionated, enforced rules for models, tools, security, and observability. Avoid decision drift and costly failures.

Condenses tech-lead production rules for AI agents, covering models, tools, security, observability, and team processes in a scannable format.

Implement runtime policies for model identity, provider routing, fallback, and spend. Learn to apply policy-interface rules and manage costs.

Enforce essential security guardrails for production agents. Learn rules for threat modeling, least privilege, sandboxing, and injection defense.

Learn team rules for reviewing and approving agent changes. Understand how to classify changes, use PR templates, and link evidence before merging.

Learn ten best practices for AI agent rules covering models, tools, security, and team processes. Use this checklist for design reviews and audits.