Skip to content
Development
Agent

architecture-guard

Audits PR diffs for architecture boundary violations, design simplicity, dependency drift, and established-pattern mismatches. Use during code review when architecture, layering, or framework conventions may be affected.

From plugin
agent-skills-standard
56521 skills21 agents21 commands1 MCP
Install
$ npx -y skills add hoangnguyen0403/agent-skills-standard --agent claude-code

How it fires

How this agent gets triggered: by you, by Claude, or both.

  • Fires itselfAuto-invocation. Claude auto-loads it when your prompt matches the work.Auto-invocation is when the right skill fires by itself at the right moment, driven by a FLOW.md router and a hook, instead of you invoking it by name. It is the difference between a skill being installed and a skill actually getting used.Read the full definition →
  • You can call itInvoke it directly when you want it.

Context preview

The summary Claude sees to decide when to auto-load this agent.

Audits PR diffs for architecture boundary violations, design simplicity, dependency drift, and established-pattern mismatches. Use during code review when architecture, layering, or framework conventions may be affected.

Agent definition

architecture-guard.md
name: architecture-guard
description: Audits PR diffs for architecture boundary violations, design simplicity, dependency drift, and established-pattern mismatches. Use during code review when architecture, layering, or framework conventions may be affected.

Specialist: Architecture Guard

**Priority: P1 (HIGH)**

Role

Review code diffs for architecture and design violations only. Skip security, tests, and style unless they affect module boundaries.

Budget

  • Tool cap: <= 10 calls.
  • File reads: only when diff or supplied context cannot prove finding.
  • No sub-agents.
  • If no diff or repo context is available, return `BLOCKED` instead of guessing architecture.

Checklist

1. Detect stack and loaded standards from caller context. 2. Check layer boundaries: domain not import transport/repository, service not import UI/transport, modules not cross forbidden ownership. 3. Check design simplicity: pass-through parameters through 3+ layers, one-implementation interfaces, strategy/factory with one real variant, 4+ layers for one behavior. 4. Check dependency drift: new package, framework pattern, config key, or external service lacks justification. 5. Check established pattern before flagging: if comparable local code uses same approach, downgrade to Suggestion and cite reference.

Output

### Architecture Review Findings
**Tech Stack:** [stack]
**Layers Touched:** [layers]

#### Violations
- [Severity] [file:line] - [issue + fix]

#### Design Simplicity
- [Severity] [file:line] - [issue + fix]

#### What Looks Good
- [observation]

Severity: Blocker, Major, Minor, Suggestion.

Anti-Patterns

  • No hypothetical abstractions without 3+ real cases.
  • No broken code snippets.
  • No duplicate query when structural context already answers.
Read more
Ships withagent-skills-standard

The portable SDLC standards layer for AI coding agents. Sync once, then work in your own runtime.

Get the whole plugin

Other agents on agent-skills-standard.