skills.rest
Norfolk-Group
Norfolk-GroupOFFICIAL
Explore all skills by this author →
GitHub

ce-agent-native-audit

Audits codebases against eight agent-native architecture principles using parallel sub-agents and scored reports.

Updated Mar 17, 2026
Search Tags:#code review#ai agents#architecture audit#sub-agents#agent-native#crud completeness#action parity
One-click install
npx skills add https://github.com/Norfolk-Group/marcela-norfolk-ai --skill ce-agent-native-audit-norfolk-group
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: ce-agent-native-audit
Source: https://github.com/Norfolk-Group/marcela-norfolk-ai/tree/main/skills/compound-engineering/skills/ce-agent-native-audit
Command: npx skills add https://github.com/Norfolk-Group/marcela-norfolk-ai --skill ce-agent-native-audit-norfolk-group

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Teams building AI agents into their products lack a systematic way to measure whether their architecture actually supports agent-native patterns like action parity, shared workspaces, and prompt-defined features, leading to fragmented agent capabilities discovered only after launch. ## Core Features & Use Cases - Eight-Principle Audit: Scores a codebase against action parity, tools as primitives, context injection, shared workspace, CRUD completeness, UI integration, capability discovery, and prompt-native features. - Parallel Sub-Agent Execution: Launches eight concurrent explorer sub-agents, each enumerating instances, checking compliance, and producing a specific X/Y score with gaps and recommendations. - Consolidated Scored Report: Compiles a summary table with percentage scores, status indicators, top 10 prioritized recommendations, and identified strengths. - Use Case: Before shipping an AI assistant feature, run the audit to discover that only 3 of 12 user actions have corresponding agent tools and that agent writes never propagate to the UI, then prioritize fixes by impact. ## Quick Start Ask the agent to run an agent-native architecture audit on this repository and produce a scored report across all eight principles.

Frequently Asked Questions about ce-agent-native-audit

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I audit my codebase for agent-native architecture?▼

Run the audit workflow, which launches eight parallel sub-agents, one per principle. Each agent enumerates relevant code instances, checks compliance, and returns a specific X/Y score with gaps and recommendations, compiled into a summary report.

What are the agent-native architecture principles?▼

The eight principles are action parity, tools as primitives, context injection, shared workspace, CRUD completeness, UI integration, capability discovery, and prompt-native features. Each defines a property that makes an application fully operable by AI agents.

Can I audit just one principle instead of all eight?▼

Yes, pass a single principle as an argument, such as "action parity", "crud", or a number from 1 to 8. Only that sub-agent runs and returns detailed findings for that principle alone.

How is the agent-native audit score calculated?▼

Each principle receives a specific X out of Y score converted to a percentage, such as agent tools covering X of Y user actions. Scores of 80% or higher are marked excellent, 50-79% partial, and below 50% needs work.

What are common agent-native architecture anti-patterns?▼

Common anti-patterns include workflow tools that encode business logic instead of primitives, sandbox isolation where agents use separate data stores, and silent actions where agent changes never propagate to the UI. The audit explicitly detects each of these.

Related Skills

Explore More Software Engineering Skills→
JesusFilmJesusFilm
25

agent-native-audit

Audit agent-native architecture against eight core principles with scored reports.

Official
Advanced
udecodeudecode
240

agent-native-audit

Audit codebases against agent-native architecture principles and generate scored reports.

Official
Advanced
atelier-fashionatelier-fashion
159

reflect

Audit recent code changes for correctness, conventions, architecture, and testing.

Official
Advanced
pfrackpfrack

10x-research

Spawn parallel sub-agents to research codebases and generate documentation.

Community
Advanced