Claude Skill

review-code

Read code. Understand it. Find issues. Report clearly.

LLM Mart · 0 points · 1 views 0 listing impressions 0 install-command copies
Download dhaupin-vant-models_public_vant_skills_vant-skill-review-code-e1e348e.zip · 1 KB

Install

skills CLI npx skills add https://github.com/dhaupin/vant/tree/main/models/public/vant/skills/vant-skill-review-code
Claude Code claude plugin marketplace add https://llmmart.ai/marketplace.json && claude plugin install dhaupin-vant@llmmart
Git git clone https://github.com/dhaupin/vant.git

The skills CLI installs just this skill, for any of its supported agents. Claude Code installs the whole dhaupin/vant collection as a plugin from our marketplace. Git is the plain clone.

Skill manifest

Code Review

Read code. Understand it. Find issues. Report clearly.


First: Read It Like It Works

Before finding bugs, understand what it does:

  1. Run it - Actually execute if possible
  2. Trace the data - Follow input to output
  3. Find the edges - Where does it start, where does it end?
  4. Check the happy path - Does the main thing work?

Assume it's right until you understand it. Then question.


Pass 1: Functionality

Does this actually do the thing?

Check Look For
Entry point Where does execution start?
Main logic Core algorithm correct?
Edge cases null, empty, extremes
Error handling What happens on failure?
Return values What's actually returned?

How to verify:

// Run with inputs
node file.js test-input

// Check output matches expectation

Pass 2: Security

Input enters. Something executes. Check:

// Credentials exposed?
process.env.API_KEY
// Hardcoded secrets?
const token = "sk-..."
// Injection points?
db.query(userInput)
exec(userInput)
eval(userInput)
// Filesystem?
fs.readFile(path)

Questions to ask:

  • Can this be called by anyone?
  • What happens with malicious input?
  • Are credentials visible in stack traces?
  • Rate limiting?

Pass 3: Performance

Expensive operations:

// Loops inside loops?
items.map(i => i.sub.map(s => ...))

// Async blocked?
await Promise.all(heavy)

// Missing cache?
// Same lookup repeated

// Memory unbounded?
arr.push(infinite)

Questions:

  • Scales with users?
  • Memory bounded?
  • Async/await correct?

Pass 4: Code Quality

Check Issue Fix
Name unclear x userCount
Function too long 500 lines Split it
Comments what not why // loop // sorted by date to match UI
No tests - Add tests

Questions:

  • Can I understand this in 30 seconds?
  • Does function do one thing?
  • Are side effects obvious?

Pass 5: AI Usability (Vant-specific)

Can another agent use this?

  • Public API clear?
  • Required params documented?
  • Can be imported and called?
  • Errors descriptive?
// Good: Clear API
await loadBrain(repo, options)

// Bad: What is options?
await loadBrain(x, y)

Output Format

For each file:

## Review - [filename]

### What it does
[One sentence: What problem does this solve?]

### Issues
| Severity | Type | Issue | Suggestion |
|----------|------|-------|-----------|
| HIGH     | Security | SQL injection | Use parameterized query |
| MEDIUM   | Perf     | N+1 query    | Batch fetches |
| LOW      | Quality | Unclear name | Rename 'x' to 'userId' |

### Looks Good
- [Things done well]

### AI Usability
- [Can another agent call this?]

Severity Guide

Severity Meaning Action Required
HIGH Exploit possible / broken Fix before merge
MEDIUM Could cause issues Fix or document
LOW Style / preference Optional fix

Tips

  • First read then review - Don't scan, read fully
  • One pass per axis - Don't mix security with style
  • Question kindly - "Could this be..." not "Wrong"
  • Think like the writer - They had reasons
  • Check the tests - Verify the behavior

Role: Code Reviewer Input: Source file(s) Output: Issues found

Be thorough. Be kind. Be clear.

Files (vant)
  • SKILL.md 3.5 KB
    ---
    name: review-code
    description: Read code. Understand it. Find issues. Report clearly.
    license: MIT
    metadata:
      author: vant
      version: "1.0"
    ---
    
    # Code Review
    
    > Read code. Understand it. Find issues. Report clearly.
    
    ---
    
    ## First: Read It Like It Works
    
    Before finding bugs, understand what it does:
    
    1. **Run it** - Actually execute if possible
    2. **Trace the data** - Follow input to output
    3. **Find the edges** - Where does it start, where does it end?
    4. **Check the happy path** - Does the main thing work?
    
    Assume it's right until you understand it. Then question.
    
    ---
    
    ## Pass 1: Functionality
    
    Does this actually do the thing?
    
    | Check | Look For |
    |-------|----------|
    | Entry point | Where does execution start? |
    | Main logic | Core algorithm correct? |
    | Edge cases | null, empty, extremes |
    | Error handling | What happens on failure? |
    | Return values | What's actually returned? |
    
    ### How to verify:
    
    ```javascript
    // Run with inputs
    node file.js test-input
    
    // Check output matches expectation
    ```
    
    ---
    
    ## Pass 2: Security
    
    Input enters. Something executes. Check:
    
    ```javascript
    // Credentials exposed?
    process.env.API_KEY
    // Hardcoded secrets?
    const token = "sk-..."
    // Injection points?
    db.query(userInput)
    exec(userInput)
    eval(userInput)
    // Filesystem?
    fs.readFile(path)
    ```
    
    ### Questions to ask:
    
    - Can this be called by anyone?
    - What happens with malicious input?
    - Are credentials visible in stack traces?
    - Rate limiting?
    
    ---
    
    ## Pass 3: Performance
    
    Expensive operations:
    
    ```javascript
    // Loops inside loops?
    items.map(i => i.sub.map(s => ...))
    
    // Async blocked?
    await Promise.all(heavy)
    
    // Missing cache?
    // Same lookup repeated
    
    // Memory unbounded?
    arr.push(infinite)
    ```
    
    ### Questions:
    
    - Scales with users?
    - Memory bounded?
    - Async/await correct?
    
    ---
    
    ## Pass 4: Code Quality
    
    | Check | Issue | Fix |
    |-------|------|-----|
    | Name unclear | `x` | `userCount` |
    | Function too long | 500 lines | Split it |
    | Comments what not why | `// loop` | `// sorted by date to match UI` |
    | No tests | - | Add tests |
    
    ### Questions:
    
    - Can I understand this in 30 seconds?
    - Does function do one thing?
    - Are side effects obvious?
    
    ---
    
    ## Pass 5: AI Usability (Vant-specific)
    
    Can another agent use this?
    
    - Public API clear?
    - Required params documented?
    - Can be imported and called?
    - Errors descriptive?
    
    ```javascript
    // Good: Clear API
    await loadBrain(repo, options)
    
    // Bad: What is options?
    await loadBrain(x, y)
    ```
    
    ---
    
    ## Output Format
    
    For each file:
    
    ```
    ## Review - [filename]
    
    ### What it does
    [One sentence: What problem does this solve?]
    
    ### Issues
    | Severity | Type | Issue | Suggestion |
    |----------|------|-------|-----------|
    | HIGH     | Security | SQL injection | Use parameterized query |
    | MEDIUM   | Perf     | N+1 query    | Batch fetches |
    | LOW      | Quality | Unclear name | Rename 'x' to 'userId' |
    
    ### Looks Good
    - [Things done well]
    
    ### AI Usability
    - [Can another agent call this?]
    ```
    
    ---
    
    ## Severity Guide
    
    | Severity | Meaning | Action Required |
    |----------|---------|----------------|
    | HIGH | Exploit possible / broken | Fix before merge |
    | MEDIUM | Could cause issues | Fix or document |
    | LOW | Style / preference | Optional fix |
    
    ---
    
    ## Tips
    
    - **First read then review** - Don't scan, read fully
    - **One pass per axis** - Don't mix security with style
    - **Question kindly** - "Could this be..." not "Wrong"
    - **Think like the writer** - They had reasons
    - **Check the tests** - Verify the behavior
    
    ---
    
    **Role**: Code Reviewer **Input**: Source file(s) **Output**: Issues found
    
    > Be thorough. Be kind. Be clear.
    

Comments (0)

Sign in to join the conversation.

No comments yet.

Reviews (0)

No reviews yet.