agent-native-audit

An agent-native architecture audit evaluates a codebase against eight principles like action parity and context injection, identifying gaps between user and agent capabilities. The audit uses parallel sub-agents to generate a structured, per-principle scored report.

To audit your codebase for agent-native architecture, invoke the skill to enumerate user actions, tools, context, and data stores. It launches parallel sub-agents for each principle and compiles a final SCORE report with actionable recommendations.

The audit evaluates eight principles: Action Parity, Tools as Primitives, Context Injection, Shared Workspace, CRUD Completeness, UI Integration, Capability Discovery, and Prompt-Native Features. Each principle receives a specific score from a dedicated sub-agent.

Yes, the agent-native architecture audit produces a structured final summary with an overall score and a matrix of per-principle results. This format is designed specifically for quick leadership reviews and actionable improvement tracking.

Yes, the agent-native architecture audit generates concrete gaps and actionable recommendations for improvement. After evaluating each of the eight principles, it compiles these steps into a structured SCORE report to guide your codebase enhancements.

The best way to identify agent capability gaps is running an agent-native architecture audit. It enumerates user actions and UI interactions, then uses parallel sub-agents to score principles like CRUD completeness and capability discovery.