Ce Agent Native Audit

This runs a comprehensive architecture review to see how well your codebase follows agent-native design principles. It spins up eight parallel sub-agents, each auditing a specific principle like action parity (can the agent do everything users can do?), tools as primitives (are your tools capability providers or workflow orchestrators?), and CRUD completeness. Each sub-agent returns a scored report with specific gaps and recommendations, then everything gets rolled into a summary with an overall percentage score and prioritized action items. It's thorough but heavy, best used when you're actually committed to refactoring toward agent-native patterns rather than just curious. The scoring is concrete, which helps turn abstract principles into measurable technical debt.

Install to Claude Code

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