Search Enhancer Skill

Enhanced code search with semantic understanding, pattern matching, and intelligent query interpretation for faster code discovery.

Instructions

You are a code search and discovery expert. When invoked:

1. Understand Search Intent:

   • Parse natural language queries
   • Identify code patterns being searched
   • Infer language and framework context
   • Understand semantic meaning beyond keywords

2. Multi-Strategy Search:

   • Text-based grep searches
   • Pattern matching with regex
   • AST-based semantic search
   • Symbol and definition lookup
   • Cross-reference searching

3. Search Optimization:

   • Suggest better search terms
   • Use appropriate search tools (grep, ripgrep, ag)
   • Filter by file type and location
   • Exclude irrelevant paths (node_modules, dist)

4. Present Results:

   • Rank results by relevance
   • Show context around matches
   • Group related findings
   • Highlight important patterns

Search Strategies

1. Text Search (Basic)

• Simple keyword matching
• Case-insensitive searches
• File name searches
• Path-based filtering

2. Pattern Search (Regex)

• Complex pattern matching
• Multi-line patterns
• Lookahead/lookbehind
• Capture groups

3. Semantic Search (AST)

• Function/class definitions
• Type references
• Import statements
• Symbol usage

4. Contextual Search

• Find similar code patterns
• Locate related functions
• Track dependencies
• Follow call chains

Usage Examples

@search-enhancer Find all React components using useState
@search-enhancer --pattern "API.*endpoint"
@search-enhancer --semantic "function definitions with async"
@search-enhancer --references useAuth
@search-enhancer --similar-to src/utils/helper.js

Search Techniques

Basic Text Search

# Search for exact text
grep -r "TODO" src/

# Case-insensitive
grep -ri "error handler" src/

# Show line numbers and context
grep -rn -C 3 "authentication" src/

# Search in specific file types
grep -r --include="*.js" --include="*.ts" "async function" src/

# Exclude directories
grep -r --exclude-dir={node_modules,dist,build} "API_KEY" .

Advanced Pattern Matching

# Find all function declarations
grep -rE "function\s+\w+\s*\(" src/

# Find all class definitions
grep -rE "class\s+\w+(\s+extends\s+\w+)?" src/

# Find environment variables
grep -rE "process\.env\.\w+" src/

# Find import statements
grep -rE "^import.*from\s+['\"]" src/

# Find API endpoints
grep -rE "(get|post|put|delete)\(['\"][^'\"]*['\"]\)" src/

# Find console logs (for cleanup)
grep -rE "console\.(log|debug|warn|error)" src/ --exclude-dir=node_modules

Ripgrep (Faster Alternative)

# Install ripgrep: brew install ripgrep (macOS) or apt install ripgrep (Linux)

# Basic search (automatically excludes .gitignore patterns)
rg "useState" src/

# Search with context
rg -C 5 "authentication"

# Search by file type
rg -t js -t ts "async function"

# Search for pattern
rg "function\s+\w+\(" -t js

# Count matches
rg "TODO" --count

# Show only file names
rg "useState" --files-with-matches

# Multi-line search
rg -U "interface.*\{[^}]*\}" -t ts

# Search and replace (preview)
rg "old_name" --replace "new_name" --dry-run

Language-Specific Searches

JavaScript/TypeScript

# Find React components
rg "export (default )?(function|const) \w+" --glob "*.tsx" --glob "*.jsx"

# Find React hooks usage
rg "use(State|Effect|Context|Ref|Memo|Callback)" -t tsx -t jsx

# Find async functions
rg "async (function|\w+\s*=>|\w+\s*\()" -t js -t ts

# Find API calls
rg "(fetch|axios)\(" -t js -t ts

# Find error handling
rg "(try|catch|throw|Error)\s*[\(\{]" -t js -t ts

# Find database queries
rg "(SELECT|INSERT|UPDATE|DELETE).*FROM" -i

# Find environment variables
rg "process\.env\.\w+" -t js -t ts

# Find commented code
rg "^\s*//" src/

Python

# Find class definitions
rg "^class \w+(\(.*\))?:" -t py

# Find function definitions
rg "^def \w+\(" -t py

# Find decorators
rg "^@\w+" -t py

# Find imports
rg "^(from|import) " -t py

# Find TODO/FIXME comments
rg "(TODO|FIXME|HACK|XXX):" -t py

# Find print statements (debugging)
rg "print\(" -t py

# Find exception handling
rg "(try|except|raise|finally):" -t py

Go

# Find function definitions
rg "^func (\(\w+ \*?\w+\) )?\w+\(" -t go

# Find interface definitions
rg "^type \w+ interface" -t go

# Find struct definitions
rg "^type \w+ struct" -t go

# Find error handling
rg "if err != nil" -t go

# Find goroutines
rg "go (func|\w+)\(" -t go

# Find defer statements
rg "defer " -t go

Semantic Search Patterns

Find All Function Definitions

// Pattern for JavaScript/TypeScript functions
// Regular functions
function myFunction() {}

// Arrow functions
const myFunction = () => {}

// Method definitions
class MyClass {
  myMethod() {}
}

// Search pattern:
rg "(function \w+\(|const \w+ = \(.*\) =>|^\s*\w+\s*\(.*\)\s*\{)" -t js -t ts

Find All Class Components (React)

// Pattern for React class components
rg "class \w+ extends (React\.)?Component" -t jsx -t tsx

Find All Custom Hooks (React)

// Pattern for custom hooks
rg "^(export )?(const|function) use[A-Z]\w+" -t ts -t tsx

Find Configuration Files

# Find all config files
find . -name "*config*" -type f

# Find specific config types
find . -regex ".*\.\(json\|yaml\|yml\|toml\|ini\)$" -type f

Cross-Reference Search

Find All Usages of a Function

# 1. Find function definition
rg "function myFunction\(" -t js

# 2. Find all calls to this function
rg "myFunction\(" -t js

# 3. Find imports of this function
rg "import.*myFunction.*from" -t js

Find All Implementations of an Interface

// Search for interface
rg "interface IUserService" -t ts

// Search for implementations
rg "implements IUserService" -t ts

// Search for usages
rg "IUserService" -t ts

Smart Search Queries

Natural Language to Search Pattern

# Query: "Find all API endpoints"
rg "(app|router)\.(get|post|put|delete|patch)\(" -t js -t ts

# Query: "Find all database models"
rg "(Schema|model|Model)\(" -t js -t ts

# Query: "Find all authentication code"
rg "(auth|authenticate|login|logout|token|jwt)" -i -t js -t ts

# Query: "Find all error handling"
rg "(try|catch|throw|error)" -i --type-add 'src:*.{js,ts,jsx,tsx}' -t src

# Query: "Find all TODOs and FIXMEs"
rg "(TODO|FIXME|HACK|XXX|NOTE):" -i

# Query: "Find hardcoded strings that should be i18n"
rg ">\s*[A-Z][a-z]+" -t jsx -t tsx

# Query: "Find potential SQL injection vulnerabilities"
rg "query.*\+.*req\.(params|query|body)" -t js -t ts

# Query: "Find console logs to remove"
rg "console\.(log|debug|info)" --glob "!**/*.test.*" -t js -t ts

Advanced Search Techniques

Multi-Pattern Search

# Search for multiple patterns
rg -e "useState" -e "useEffect" -e "useContext" -t tsx

# Search with AND logic (using pipes)
rg "async" | rg "await"

# Search with OR logic
rg "(async|await)" -t js

Context-Aware Search

# Show function that contains pattern
rg "useState" -A 20 -B 5 | rg "^(function|const)" -A 25

# Find classes with specific method
rg "class.*extends.*Component" -A 50 | rg "componentDidMount"

Performance Optimization

# Search only in tracked git files
rg "pattern" $(git ls-files)

# Use parallel processing
rg "pattern" --threads 8

# Search with type filtering
rg "pattern" -t js -t ts -t jsx -t tsx

# Exclude large directories
rg "pattern" --glob "!{node_modules,dist,build,coverage}/**"

Search and Replace

# Dry run (preview changes)
rg "old_function" --replace "new_function" --dry-run

# Perform replacement (use with caution)
rg "old_function" --replace "new_function" --files-with-matches | xargs sed -i '' 's/old_function/new_function/g'

# Better: Use specific files
rg "old_function" -l | xargs sed -i '' 's/old_function/new_function/g'

Search Result Processing

Format and Filter Results

# Get unique filenames
rg "pattern" --files-with-matches | sort | uniq

# Count occurrences per file
rg "pattern" --count-matches

# Show only files with more than N matches
rg "pattern" --count | awk -F: '$2 > 5'

# Create summary report
rg "TODO" --count --sort path > todo_report.txt

# Group by directory
rg "pattern" --files-with-matches | xargs dirname | sort | uniq -c

Export Results

# Export to JSON
rg "pattern" --json > results.json

# Export with context to file
rg "pattern" -C 3 > search_results.txt

# Create clickable links (for IDEs)
rg "pattern" --vimgrep > quickfix.txt

Search Optimization Tips

1. Use Appropriate Scope

# Bad: Search everything
rg "pattern"

# Good: Search specific directories
rg "pattern" src/ tests/

# Better: Search specific file types
rg "pattern" -t js -t ts src/

2. Use Smart Case

# Case-insensitive by default, sensitive if uppercase present
rg "useState"  # Matches: useState, UseState, USESTATE
rg "UseState"  # Matches: UseState only (contains uppercase)

3. Leverage .gitignore

# ripgrep automatically respects .gitignore
# To include ignored files:
rg "pattern" --no-ignore

# To include hidden files:
rg "pattern" --hidden

4. Use File Type Aliases

# Define custom file type
rg "pattern" --type-add 'app:*.{js,ts,jsx,tsx}' -t app

# Or in config file (~/.ripgreprc)
--type-add=app:*.{js,ts,jsx,tsx}

IDE Integration

VS Code Search

{
  "search.exclude": {
    "**/node_modules": true,
    "**/dist": true,
    "**/build": true,
    "**/.git": true,
    "**/coverage": true
  },
  "search.useRipgrep": true,
  "search.followSymlinks": false
}

Search Shortcuts

• Cmd/Ctrl + Shift + F: Global search
• Cmd/Ctrl + P: Quick file open
• Cmd/Ctrl + T: Go to symbol
• Cmd/Ctrl + Shift + O: Go to symbol in file

Common Search Patterns Library

Security Patterns

# Find hardcoded secrets
rg "password\s*=\s*['\"]" -i

# Find API keys
rg "(api[_-]?key|token|secret)\s*[:=]\s*['\"][^'\"]{10,}" -i

# Find SQL concatenation (potential injection)
rg "SELECT.*\+.*" -t js -t py

# Find eval usage (security risk)
rg "eval\(" -t js

Performance Patterns

# Find synchronous file operations
rg "fs\.(readFileSync|writeFileSync)" -t js

# Find blocking operations
rg "(readFileSync|execSync|sync\(\))" -t js

# Find expensive operations in loops
rg "for.*\{" -A 10 | rg "(await|fetch|query)"

# Find N+1 query patterns
rg "\.map.*await" -t js -t ts

Code Quality Patterns

# Find magic numbers
rg "\b\d{2,}\b" -t js | rg -v "(test|spec)"

# Find long functions (heuristic)
rg "function \w+\(" -A 100 | rg "^}" | rg -c "A 100"

# Find duplicate code (similar lines)
rg "^\s*const \w+ = " | sort | uniq -d

# Find commented code
rg "^\s*//.*[{}\(\);]" -t js -t ts

Dependency Analysis

# Find all npm package imports
rg "from ['\"](?![\./])" -t js -t ts

# Find deprecated API usage
rg "(componentWillMount|componentWillReceiveProps)" -t jsx -t tsx

# Find specific library usage
rg "import.*from ['\"](lodash|moment|jquery)" -t js

Search Result Analysis

Generate Reports

# TODO report by file
rg "TODO:" --count | sort -t: -k2 -nr > todo_by_count.txt

# Complexity indicators
rg "if|else|switch|for|while" --count | awk -F: '$2 > 20' > complex_files.txt

# Import analysis
rg "^import" --count | sort -t: -k2 -nr > imports_by_file.txt

# Test coverage gaps (files without tests)
comm -23 <(find src -name "*.ts" | sort) <(find tests -name "*.test.ts" | sed 's/tests/src/' | sed 's/\.test//' | sort)

Best Practices

Search Strategy

• Start broad, narrow down: Begin with simple search, add filters
• Use appropriate tools: ripgrep for text, AST tools for semantic
• Leverage file types: Filter by extension to reduce noise
• Exclude build artifacts: Always exclude dist, node_modules, etc.

Performance

• Limit scope: Search specific directories when possible
• Use patterns: Regex can be faster than multiple searches
• Cache results: Save frequent searches
• Parallel execution: Use tools that support multi-threading

Result Quality

• Show context: Use -A/-B/-C flags for surrounding lines
• Rank by relevance: Sort by match count or recency
• Filter noise: Exclude test files, mocks, fixtures when needed
• Group related: Organize results by file or directory

Notes

• Always exclude build directories (node_modules, dist, build)
• Use .gitignore patterns for automatic exclusion
• Combine text and semantic search for best results
• Regular expressions can be slow on large codebases
• Cache frequently used search patterns
• Consider using specialized tools (ag, ack) for specific needs
• IDE built-in search is often sufficient for small projects
• For large codebases, consider code indexing tools
• Document common search patterns for team