Agent-Native Architecture Reviewer
Audits application code and design to ensure AI agents have full functional parity and context awareness alongside human users.
The Agent-Native Architecture Reviewer is a specialized diagnostic tool designed to transition applications into 'agent-native' environments where AI agents are first-class citizens. It systematically reviews codebases to ensure that every action available to a human user in the UI has a corresponding tool for an agent, and that agents have access to the same runtime context as users. By identifying anti-patterns like context starvation and sandbox isolation, this skill helps developers build robust systems where agents and humans can collaborate seamlessly in a shared workspace using flexible, primitive-based tools.