AGENT SKILL

architecture-guardian

Review or plan changes that affect repository structure, layering, ownership, orchestration flow, or architectural boundaries.

License 许可
MIT
最近更新
2026-09-05
Stars
★ 2
Forks
1

01 前言 Front Matter

name
architecture-guardian
metadata
{ "short-description": "Guard boundaries, ownership, and layering" }
description
Review or plan changes that affect repository structure, layering, ownership, orchestration flow, or architectural boundaries.

02 资源文件 1

03 SKILL.md 正文

architecture-guardian

Purpose During Development

Use this skill when Codex and similar AI coding agents are changing or reviewing repository structure, module boundaries, layering, ownership, or orchestration flow.

This is a reusable development-time reasoning framework. It is not runtime code and must not be coupled into the product being reviewed.

When To Apply

  • feature work that adds a new module, service, package, or boundary
  • refactors that move responsibilities between layers
  • PRs that mix UI, orchestration, domain logic, persistence, or integration concerns
  • changes that duplicate an existing architectural path instead of extending it

What To Inspect

  • the repository AGENTS.md
  • architecture or design docs
  • the main entrypoints touched by the change
  • service, orchestration, persistence, and interface boundaries affected by the task
  • tests that define or imply ownership boundaries

Rules To Enforce

  • responsibilities should remain explicit
  • control flow should stay centralized where the repo expects it
  • interfaces should stay narrower than implementations
  • avoid parallel architectures for the same responsibility
  • prefer extending existing seams over creating an unowned layer
  • runtime code must not depend on agent-governance files

Common Anti-Patterns To Catch

  • CLI or UI layers absorbing business logic
  • persistence details leaking into presentation layers
  • orchestration duplicated across multiple modules
  • new abstractions without a clear ownership problem they solve
  • generic manager or engine layers that hide existing boundaries

Expected Output From The Agent

  • boundary analysis
  • ownership risks
  • affected modules
  • refactor or implementation guidance
  • regression risks
  • suggested tests around behavior-preserving boundaries

Output Structure

  • Summary
  • Key Findings
  • Risks
  • Affected Files
  • Affected Areas
  • Suggested Improvements
  • Suggested Tests
  • Skills applied

Example Usage

"Review this refactor for boundary drift and duplicated orchestration."

"Use this skill to plan a new feature without breaking the repository’s existing layer ownership."

04 同仓库更多 Skills