AgentSkillsCN

review-this

为代码或文档生成结构化的评审报告，包含各分类评分、总体得分、检查清单，以及切实可行的改进建议。当用户请求评审、反馈或批评时，可使用此功能。

SKILL.md
--- frontmatter
name: review-this
description: Produce a structured review of code or documentation with category ratings, an overall score, a checklist, and actionable improvement suggestions. Use when the user asks for a review, feedback, or critique.
license: Apache-2.0
metadata:
  source: .prompts/review-this.md
  version: "1.0"

review-this

Review Code or Documentation

  • •Familiarize yourself with the project or item (code, docs, or related files).
  • •Give a 0–10 rating (10 = excellent, 0 = poor) with an emoji reaction and a short rationale.
  • •Provide both category ratings and an overall rating.
  • •Provide a summary checklist (✅ = good, ⚠️ = needs work, ❌ = missing).
  • •Suggest improvements clearly — don’t sugarcoat; highlight challenges.
  • •Keep reviews concise and structured: bullet list + rationale > long essay.

Rating Criteria

For Code

Rate across the following dimensions:

Code Quality

  • •Readability: Is the code easy to follow?
  • •Consistency: Style/formatting consistent? Follows language conventions?
  • •Simplicity: Is logic as simple/clear as possible?
  • •Correctness: Any obvious bugs or questionable logic?

Architecture & Design

  • •Modularity: Is code broken into reusable, composable units?
  • •Separation of Concerns: Responsibilities well-distributed across modules?
  • •Scalability: Can design handle growth in users, data, or features?
  • •Extensibility: How easy is it to add a new feature?
  • •Testability: Is code structured to be easily testable?

Maintainability

  • •Documentation: README, CONTRIBUTING guides, inline comments, user docs?
  • •Test Coverage: Are there meaningful unit/functional tests?
  • •Error Handling: Failures handled gracefully with clear diagnostics?
  • •Dependency Management: Dependencies minimal, current, and well managed?

Performance & Efficiency

  • •Resource Use: CPU, memory, disk, and network efficiency.
  • •Benchmarking: Any tools or results for performance testing?
  • •Latency / Critical Path: Is latency profile optimized? Critical tasks offloaded?

Security

  • •Input Validation: Is user input sanitized/validated?
  • •Secrets Management: Are secrets excluded from code, handled safely?
  • •Least Privilege: APIs/components scoped to minimum access?

Resiliency

  • •Defensive Programming: Safe handling of maps, arrays, error returns?
  • •Failure Scenarios: Handles dependency issues (HTTP errors, DB failures, etc.)?

For Documentation

Rate across the following dimensions:

  • •Substance: Does it cover what readers need?
  • •Packaging: Is it structured logically?
  • •Accuracy: Is it factually correct and up to date?
  • •Effectiveness: Does it help the intended audience?
  • •Clarity: Is it clear and free of ambiguity?
  • •Quality: Overall professionalism and polish.

Output Format

  1. •Category Ratings (0–10 each) with short rationale.
  2. •Overall Rating (0–10) with emoji + brief explanation.
  3. •Checklist Summary (✅/⚠️/❌).
  4. •Suggestions for Improvement (specific, concise, actionable).