ast-grep-code-analysis（ast grep 代码分析）

AST-Grep Code Analysis

[!NOTE] This skill requires that as-grep is installed and configured in your development environment. If it's not installed we can use mise -g --pin ast-grep

Overview

AST-Grep Code Analysis uses Abstract Syntax Tree pattern matching to systematically identify code issues, replacing manual line-by-line inspection with structural pattern recognition.

Core principle: Code structure reveals more than surface reading - AST patterns expose hidden relationships, security vulnerabilities, and architectural issues that manual inspection misses.

When to Use

digraph when_to_use {
    "Need to analyze code?" [shape=diamond];
    "Complex/nested structure?" [shape=diamond];
    "Security review needed?" [shape=diamond];
    "Performance analysis?" [shape=diamond];
    "Use ast-grep patterns" [shape=box];
    "Manual review sufficient" [shape=box];

    "Need to analyze code?" -> "Complex/nested structure?" [label="yes"];
    "Complex/nested structure?" -> "Security review needed?" [label="yes"];
    "Security review needed?" -> "Performance analysis?" [label="yes"];
    "Performance analysis?" -> "Use ast-grep patterns" [label="yes"];
    "Complex/nested structure?" -> "Manual review sufficient" [label="no"];
    "Security review needed?" -> "Manual review sufficient" [label="no"];
    "Performance analysis?" -> "Manual review sufficient" [label="no"];
}

Use when:

• Code has nested functions, complex control flow, or multiple abstraction layers
• Security review required (authentication, authorization, data handling)
• Performance analysis needed (React hooks, loops, async patterns)
• Large codebase where manual inspection is impractical
• Need to identify patterns across multiple files
• Time pressure requires systematic approach over ad-hoc analysis

Do NOT use when:

• Simple, straightforward code (< 50 lines)
• Single-file utilities with obvious structure
• When quick glance is sufficient for the task

Core Pattern

Before (Manual Inspection):

// Agent manually reads line by line
if (data[i].admin) {
  userObj.token = generateToken(data[i].id); // "This looks insecure"
}

After (AST Pattern Matching):

# ast-grep rule: insecure-token-generation
rule:
  pattern: |
    function $FUNC($ARGS) {
      const secret = $SECRET;
      return btoa(JSON.stringify($PAYLOAD) + '.' + $SECRET);
    }
  meta:
    severity: ERROR
    message: "Hardcoded secret in token generation"

Quick Reference

Implementation

Installation and Setup

# Install ast-grep
npm install -g @ast-grep/cli

# Initialize configuration
ast-grep init

# Create rules directory
mkdir -p sg-rules/security sg-rules/performance sg-rules/structure

Essential Security Patterns

Hardcoded Secrets Detection:

# sg-rules/security/hardcoded-secrets.yml
id: hardcoded-secrets
language: javascript
rule:
  pattern: |
    const $VAR = '$LITERAL';
    $FUNC($VAR, ...)
  meta:
    severity: ERROR
    message: "Potential hardcoded secret detected"

Insecure Token Generation:

# sg-rules/security/insecure-tokens.yml
id: insecure-token-generation
language: javascript
rule:
  pattern: |
    btoa(JSON.stringify($OBJ) + '.' + $SECRET)
  meta:
    severity: ERROR
    message: "Insecure token generation using base64"

Performance Pattern Detection

React Hook Dependencies:

# sg-rules/performance/react-hook-deps.yml
id: react-hook-dependency-array
language: typescript
rule:
  pattern: |
    useEffect(() => {
      $BODY
    }, [$FUNC])
  meta:
    severity: WARNING
    message: "Function dependency in useEffect may cause infinite re-renders"

Missing useCallback Optimization:

# sg-rules/performance/missing-use-callback.yml
id: missing-use-callback
language: typescript
rule:
  pattern: |
    const $FUNC = ($ARGS) => {
      $BODY
    };
  inside:
    kind: function_declaration
    has:
      kind: arrow_function
  meta:
    severity: INFO
    message: "Consider wrapping function in useCallback for optimization"

Structural Analysis Patterns

Deep Nesting Detection:

# sg-rules/structure/deep-nesting.yml
id: deep-nesting
language: javascript
rule:
  any:
    - pattern: |
        if ($COND1) {
          if ($COND2) {
            if ($COND3) {
              $BODY
            }
          }
        }
    - pattern: |
        for ($INIT) {
          for ($INIT2) {
            for ($INIT3) {
              $BODY
            }
          }
        }
  meta:
    severity: WARNING
    message: "Deep nesting detected - consider refactoring"

Running Analysis

# Run all security rules
ast-grep run -r sg-rules/security/

# Run performance analysis on React components
ast-grep run -r sg-rules/performance/ --include="*.tsx,*.jsx"

# Generate comprehensive report
ast-grep run -r sg-rules/ --format=json > analysis-report.json

# Interactive analysis
ast-grep run -r sg-rules/ --interactive

Common Mistakes

Real-World Impact

Before AST Analysis:

• Manual code review: 2-3 hours for medium codebase
• Missed security vulnerabilities: 40-60%
• Inconsistent analysis between reviewers
• No systematic approach to pattern detection

After AST Analysis:

• Automated pattern detection: 5-10 minutes
• Security vulnerability detection: 90%+
• Consistent, repeatable analysis
• Comprehensive coverage of known anti-patterns

Example Results:

$ ast-grep run -r sg-rules/
src/components/UserProfile.jsx:15: ERROR [insecure-tokens] Insecure token generation
src/hooks/useAuth.js:8: ERROR [hardcoded-secrets] Potential hardcoded secret
src/components/UserProfile.jsx:23: WARNING [react-hook-deps] Function dependency may cause re-renders
src/utils/processData.js:45: WARNING [deep-nesting] Deep nesting detected

Found 4 issues (2 errors, 2 warnings)

Integration Workflow

1. Setup: Create rule sets for security, performance, structure
2. Baseline: Run analysis on existing codebase to establish patterns
3. Iterate: Refine rules based on false positives/negatives
4. Automate: Integrate into CI/CD pipeline for continuous analysis
5. Monitor: Track issue reduction over time

Required Background: Understanding of AST concepts, pattern matching, and code structure analysis. AST patterns reveal what manual inspection misses - systematic, comprehensive, and repeatable code analysis.