跳到正文
cccskills
免费在 GitHub 公开

reviewer-architecture

Performs architectural code reviews focusing on design patterns, duplication, circular dependencies, and refactoring impact. Is the design consistent with the codebase? Use when reviewing code changes for architectural concerns, code smells, or structural issues.

中文简介正在准备中。当前显示原文说明。

实测评分 未实测

查看安装方法

包含的文件(1)

  • SKILL.md4.3 KB

SKILL.md(原文)

安装前可以查看交给 AI 代理的指令内容。

Architectural Reviewer

Performs deep architectural analysis of code changes, identifying design patterns, duplication, circular dependencies, and refactoring side effects.

Review Complete When

  • Scope of affected files and modules is mapped
  • Duplication is audited (no new unnecessary copies, consolidation opportunities noted)
  • Patterns verified against codebase conventions (caller expectations, interface consistency)
  • Circular dependencies validated or ruled out for all affected modules
  • Refactoring impact assessed (stub files, side effects, progression toward goal)
  • Summary table shows status for all items
  • Overall verdict is clear ([PASS] patterns sound / [WARN] concerns noted / [FAIL] rework needed)

A review is not complete if any architectural check has an "unknown" or "not checked" status.

Instructions

1. Identify Scope

Determine which files and modules are affected by the changes. Map the import/export graph for the affected modules.

2. Check for Duplication

Search for duplicate or near-duplicate implementations across the codebase:

# Find similar function names across files
grep -rn "function getConsolidationFiles" --include="*.ts" --include="*.js"

# Find similar patterns (e.g., path.join, fs.readdirSync)
grep -rn "path.join.*map" --include="*.ts" --include="*.js"

For each duplicate found, assess:

  • Are the implementations identical or divergent?
  • Is there an intentional reason for the duplication?
  • Should they be consolidated into a shared utility?

3. Assess Architectural Patterns

Evaluate whether the changes follow existing patterns in the codebase:

  • Path resolution: Do path-building functions return full paths consistently? Check against existing patterns like discoverAgents().
  • Import structure: Is the import chain linear, or does it introduce cycles?
  • Export surface: Are exports aligned with how consumers use them?
  • Module boundaries: Do changes respect existing module responsibilities?

4. Trace Callers and Consumers

For each changed function, verify all callers:

# Find all callers of a function
grep -rn "getStructuredConsolidationFiles" --include="*.ts" --include="*.js"
grep -rn "createSession" --include="*.ts" --include="*.js"

Check:

  • Do all callers expect the same interface (e.g., full paths vs. bare filenames)?
  • Are there callers that might break with the change?
  • Are there unmet imports (functions used but not exported)?

5. Check Circular Dependencies

Build the import graph for affected modules:

shared.ts -> paths.ts (one-way)
agent-spawning.ts -> shared.ts (one-way)
dream-execution.ts -> shared.ts + agent-spawning.ts (one-way)

Flag any cycles. Circular dependencies indicate architectural coupling that should be broken.

6. Evaluate Refactoring Impact

When changes are part of a larger refactoring (e.g., extracting handlers from a monolith):

  • Are stub files created as intermediate artifacts?
  • Do new exports serve the refactoring goal?
  • Are there side effects on unrelated modules?
  • Is the refactoring progressing toward a clear architecture?

7. Report Findings

Structure the review as:

## Review

### 1. [Change Description] -- Status [PASS]/[WARN]/[FAIL]

**Evidence:**
- [Specific file:line references]
- [Pattern comparisons]
- [Caller verification]

**Verdict:** [Brief conclusion]

### 2. Architectural Concerns

#### 2a. [Issue] [WARN]/[INFO]

[Description of the concern, why it matters, and a suggestion]

### Summary

| Item | Status | Notes |
|------|--------|-------|
| [Item] | [PASS]/[WARN]/[FAIL] | [Brief note] |

**[Overall verdict]**

Use status symbols:

  • [PASS] -- Correct, no issues
  • [WARN] -- Code smell or concern worth noting (not a blocker)
  • [FAIL] -- Architectural problem that should be addressed
  • [INFO] -- Informational, known state

评价

还没有评价。欢迎分享你的使用感受。

评论

无需登录即可发表。欢迎分享使用感受、提问,或说明在哪些环境下无法运行。

还没有评论。来写第一条吧。

提交后会立即公开。请勿填写邮箱、电话等个人信息。URL 最多 2 个。

0 / 1000 字

此表单受 reCAPTCHA 保护,适用 Google 的隐私权政策和服务条款。

同一仓库的技能

Distills enduring architectural decisions and design rationale from wave orchestration plans or historical git changes into structured, tracked Architecture Decision Records (ADRs). Use pre-wave to resolve active architectural constraints or post-wave to record enduring decisions in docs/decisions/.

中文简介正在准备中。当前显示原文说明。

AndrewBrown-USDA/aigb-skills-usda72026年10月24日 更新

Create and maintain AGENTS.md files for project onboarding and grounding. Use when initialized in a new project, when project structure changes significantly, or when creating subfolder-level AGENTS.md files for complex monorepos.

中文简介正在准备中。当前显示原文说明。

AndrewBrown-USDA/aigb-skills-usda72026年10月24日 更新

Audit and update an arbitrary codebase to follow the Code.mil federal open-source licensing model (MIT with 17 U.S.C. Section 105 disclaimers and INTENT.md), including CRAN/R package DESCRIPTION setup.

中文简介正在准备中。当前显示原文说明。

AndrewBrown-USDA/aigb-skills-usda72026年10月24日 更新

Migrates ArcPy and ArcGIS Pro workflows to explicit, testable open-source Python geospatial workflows. Use when removing ArcPy dependencies, porting geoprocessing scripts, or documenting ArcGIS behavior that will not be migrated.

中文简介正在准备中。当前显示原文说明。

AndrewBrown-USDA/aigb-skills-usda72026年10月24日 更新

Adversarial tester and UX/documentation stress-tester that acts like a clumsy, literal-minded, or overwhelmed user to break systems, expose design flaws, and highlight confusing documentation. Use when the user asks to run "bull in a china shop" testing, stress-test an API/CLI/library/doc/workflow, find traps, test how a user might misuse a feature, or evaluate how documentation is interpreted.

中文简介正在准备中。当前显示原文说明。

AndrewBrown-USDA/aigb-skills-usda72026年10月24日 更新

caveman

免费

Ultra-compressed communication mode that cuts output tokens while keeping technical accuracy. Levels: lite, full, ultra and the wenyan variants. Use for /caveman, "caveman mode", "talk like caveman", "be brief" or "less tokens".

中文简介正在准备中。当前显示原文说明。

AndrewBrown-USDA/aigb-skills-usda72026年10月24日 更新

查看 AndrewBrown-USDA 的全部技能

报告此技能的问题