Agent Native Audit

This launches parallel sub-agents to audit your codebase against eight agent-native architecture principles like action parity, tools as primitives, and CRUD completeness. You get a scored report showing how well your app follows patterns where agents and users share the same workspace and capabilities. Useful if you're building AI-integrated products and want to catch architectural gaps before they become problems. The principles are opinionated but practical, things like making sure whatever a user can do through UI, an agent can do through tools. Originally from ratacat's Claude skills collection, now maintained by Every's compound engineering team.

Install to Claude Code

npx -y skills add everyinc/compound-engineering-plugin --skill agent-native-audit --agent claude-code