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
- 参考文档README.md604 B
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
test-strategy-reviewer
Assess verification scope, regression protection, and risk-based test coverage for changes under review or implementation.
security-engineer
Plan or review project security, tool reuse, dependencies, services, integrations, permissions, secrets, supply-chain risk, and security scan requirements.
scope-safety-guard
Review or plan changes that affect permissions, security boundaries, sensitive operations, external effects, or safety constraints.
project-engineer
Turn a design, idea, sketch, brief, or product concept into the simplest feasible project plan spanning design, deployment, security, operations, finance, tool reuse, and verification.
product-designer
Turn ideas, workflows, screenshots, or design briefs into credible product design direction and frontend guidance that avoids generic AI-looking UI.
playbook-installer
Guide adoption of AI-Engineering-Playbook governance files in a portable, repeatable way across repositories.