Performance Profiling & Bottleneck Detection

I'll profile your application to identify performance bottlenecks and provide optimization recommendations.

Supported Environments:

Node.js profiling (--inspect, clinic.js)
Browser performance (Chrome DevTools)
Python profiling (cProfile, line_profiler)
Bottleneck identification
Memory leak detection
Optimization recommendations

Arguments: $ARGUMENTS - optional: node|python|browser or specific file/route to profile

Token Optimization

This skill uses efficient patterns to minimize token consumption during performance profiling setup and analysis.

Optimization Strategies

1. Runtime Detection Caching (Saves 700 tokens per invocation)

Cache runtime environment and tool availability to avoid repeated detection:

CACHE_FILE=".claude/cache/performance-profile/runtime.json"
CACHE_TTL=86400  # 24 hours (tools don't change frequently)

mkdir -p .claude/cache/performance-profile

if [ -f "$CACHE_FILE" ]; then
    CACHE_AGE=$(($(date +%s) - $(stat -c %Y "$CACHE_FILE" 2>/dev/null || stat -f %m "$CACHE_FILE" 2>/dev/null)))

    if [ $CACHE_AGE -lt $CACHE_TTL ]; then
        # Use cached runtime info
        RUNTIME=$(jq -r '.runtime' "$CACHE_FILE")
        NODE_VERSION=$(jq -r '.node_version' "$CACHE_FILE")
        PYTHON_VERSION=$(jq -r '.python_version' "$CACHE_FILE")
        TOOLS_INSTALLED=$(jq -r '.tools_installed' "$CACHE_FILE")

        echo "Using cached runtime: $RUNTIME"
        echo "Tools available: $TOOLS_INSTALLED"

        SKIP_DETECTION="true"
    fi
fi

# First run: detect and cache
if [ "$SKIP_DETECTION" != "true" ]; then
    detect_runtime  # Expensive operation
    detect_tools    # Expensive operation

    # Cache results
    jq -n \
        --arg runtime "$RUNTIME" \
        --arg node "$NODE_VERSION" \
        --arg python "$PYTHON_VERSION" \
        --arg tools "$TOOLS_INSTALLED" \
        '{runtime: $runtime, node_version: $node, python_version: $python, tools_installed: $tools}' \
        > "$CACHE_FILE"
fi

Savings: 700 tokens (no repeated package.json reads, no command checks, no file system scans)

2. Early Exit for Configured Environments (Saves 85%)

If profiling already set up, provide quick summary and exit:

if [ -f ".claude/profiling/profile-node.sh" ] && [ -f ".claude/profiling/profile-$TIMESTAMP.md" ]; then
    if [ "$RECONFIGURE" != "true" ]; then
        echo "Profiling already configured"
        echo ""
        echo "Available profiling scripts:"
        ls .claude/profiling/profile-*.sh 2>/dev/null
        echo ""
        echo "Last report: $(ls -t .claude/profiling/profile-*.md | head -1)"
        echo ""
        echo "Use --reconfigure to regenerate setup"
        exit 0
    fi
fi

Savings: 85% reduction when setup exists (4,000 → 600 tokens)

3. Script Generation on Demand (Saves 70%)

Only generate profiling scripts for the detected runtime:

# Default: Only create scripts for detected runtime
case "$RUNTIME" in
    nodejs)
        create_node_profiling_script    # 800 tokens
        # Skip Python and browser scripts
        ;;
    python)
        create_python_profiling_script  # 700 tokens
        # Skip Node and browser scripts
        ;;
    browser)
        create_browser_profiling_guide  # 600 tokens
        # Skip Node and Python scripts
        ;;
esac

# Only create all scripts if explicitly requested
if [ "$GENERATE_ALL" = "true" ]; then
    create_node_profiling_script
    create_python_profiling_script
    create_browser_profiling_guide
fi

Savings: 70% (generate 1 script vs 3 scripts)

4. Sample-Based Anti-Pattern Detection (Saves 80%)

Show only first 10 issues, not exhaustive list:

# Efficient anti-pattern detection with head limit
echo "Scanning for performance anti-patterns..."

# Show only first 10 sync blocking operations
SYNC_BLOCKING=$(grep -r "readFileSync\|writeFileSync\|execSync" \
    --include="*.js" --include="*.ts" \
    --exclude-dir=node_modules --exclude-dir=dist \
    . 2>/dev/null | head -10)

if [ -n "$SYNC_BLOCKING" ]; then
    COUNT=$(echo "$SYNC_BLOCKING" | wc -l)
    echo "⚠️  Found synchronous blocking operations (showing first 10):"
    echo "$SYNC_BLOCKING" | head -5 | while read line; do
        echo "  - $line"
    done

    if [ $COUNT -gt 5 ]; then
        echo "  ... and $((COUNT - 5)) more"
    fi
fi

Savings: 80% (show samples vs exhaustive 500+ line output)

5. Grep-Based Pattern Detection (Saves 90%)

Use Grep for anti-pattern detection instead of reading all files:

# Efficient: Grep returns only matching lines
detect_performance_issues() {
    local issue_count=0

    # Check for console.log (only in src directories)
    if grep -q -r "console\.log" --include="*.{js,ts,jsx,tsx}" src/ 2>/dev/null; then
        issue_count=$((issue_count + 1))
        echo "⚠️  Issue $issue_count: console.log statements in source code"
    fi

    # Check for sync operations
    if grep -q -r "readFileSync\|writeFileSync" --include="*.{js,ts}" src/ 2>/dev/null; then
        issue_count=$((issue_count + 1))
        echo "⚠️  Issue $issue_count: Synchronous file operations detected"
    fi

    # Show count, not full lists (efficient summary)
    echo ""
    echo "Total issues detected: $issue_count"
}

Savings: 90% vs reading all files (Grep returns boolean or count)

6. Bash-Based Tool Execution (Saves 70%)

All profiling operations use pure bash (no Task agent, no Read/Edit tools):

# All operations in bash
case "$PROFILE_TYPE" in
    node)
        # Direct bash execution
        clinic doctor -- node index.js
        ;;
    python)
        # Direct bash execution
        python -m cProfile -o profile.stats main.py
        ;;
    browser)
        # Direct bash execution
        lighthouse http://localhost:3000 --output html
        ;;
esac

Savings: 70% vs using Task agents for profiling operations

7. Minimal Report Generation (Saves 60%)

Generate summary report by default, full report on demand:

REPORT_LEVEL="${REPORT_LEVEL:-summary}"

case "$REPORT_LEVEL" in
    summary)
        # Minimal report (400 tokens)
        cat > "$REPORT" << EOF
# Performance Profile Summary

**Runtime:** $RUNTIME
**Tools:** $TOOLS_INSTALLED

## Quick Start
Run profiling:
  $PROFILE_DIR/profile-$RUNTIME.sh

See full report: cat $REPORT --full
EOF
        ;;

    full)
        # Comprehensive report (2,000 tokens)
        generate_full_performance_report
        ;;
esac

Savings: 60% for default runs (summary 400 tokens vs full 2,000 tokens)

8. Tool Installation Guidance Only (Saves 95%)

Don't install tools automatically, provide installation commands:

check_tool_availability() {
    local tool="$1"

    if ! command -v "$tool" >/dev/null 2>&1; then
        echo "⚠️  $tool not installed"
        echo "   Install: $(get_install_command $tool)"
        TOOLS_MISSING="true"
    else
        echo "✓ $tool available"
    fi
}

# No actual npm install, just guidance
if [ "$TOOLS_MISSING" = "true" ]; then
    echo ""
    echo "Install missing tools to proceed with profiling"
    exit 0
fi

Savings: 95% (no npm install execution which can take 1,000+ tokens)

Cache Invalidation

Caches are invalidated when:

Runtime environment changes (different package.json, new tool installs)
24 hours elapsed (time-based for tool availability)
User runs --clear-cache flag
Major version update detected (Node.js 18 → 20)

Real-World Token Usage

Typical profiling workflow:

First-time setup: 1,500-2,500 tokens
- Runtime detection: 400 tokens
- Tool checking: 300 tokens
- Script generation (1 runtime): 800 tokens
- Summary report: 400 tokens
Cached environment: 400-800 tokens
- Cached runtime: 100 tokens (85% savings)
- Skip tool checks: 0 tokens
- Skip script regeneration: 0 tokens
- Quick summary: 300 tokens
Profiling execution: 200-600 tokens
- Execute profiling script: 200 tokens (bash operation)
- Parse results summary: 400 tokens (with --results flag)
Anti-pattern scan: 300-800 tokens
- Grep-based detection: 200 tokens
- Sample-based reporting: 100-600 tokens (show 5-10 examples)
Full report generation: 1,500-2,000 tokens
- Only when explicitly requested with --full flag

Average usage distribution:

70% of runs: Cached environment + quick summary (400-800 tokens) ✅ Most common
20% of runs: First-time setup (1,500-2,500 tokens)
10% of runs: Full analysis + report (2,000-3,000 tokens)

Expected token range: 400-2,500 tokens (60% reduction from 1,000-6,000 baseline)

Progressive Disclosure

Three levels of detail:

Default (summary): Quick setup and guidance

claude "/performance-profile"
# Shows: runtime detected, tools available, profiling commands
# Tokens: 400-800

Verbose (detailed): Anti-pattern scan + profiling scripts

claude "/performance-profile --verbose"
# Shows: full anti-pattern detection, all profiling options
# Tokens: 1,200-1,800

Full (exhaustive): Complete analysis + all runtime scripts

claude "/performance-profile --full"
# Shows: detailed report, all runtimes, complete recommendations
# Tokens: 2,000-3,000

Implementation Notes

Key patterns applied:

✅ Runtime detection caching (700 token savings)
✅ Early exit for configured environments (85% reduction)
✅ Script generation on demand (70% savings)
✅ Sample-based anti-pattern detection (80% savings)
✅ Grep-based pattern detection (90% savings)
✅ Bash-based tool execution (70% savings)
✅ Minimal report generation (60% savings)
✅ Tool guidance only, no auto-install (95% savings)

Cache locations:

.claude/cache/performance-profile/runtime.json - Runtime environment and tools
.claude/cache/performance-profile/anti-patterns.txt - Recent scan results (1 hour TTL)

Flags:

--reconfigure - Force regeneration of profiling scripts
--verbose - Show anti-pattern scan + detailed options
--full - Generate comprehensive report for all runtimes
--all-runtimes - Create scripts for Node.js + Python + Browser
--clear-cache - Force cache invalidation

Environment-specific optimizations:

Node.js: Focus on async patterns, readFileSync detection, package.json analysis
Python: Focus on list comprehensions, database queries, cProfile setup
Browser: Focus on bundle size, console.log removal, image optimization

Extended Thinking for Performance Analysis

<think> Performance profiling requires understanding: - CPU-bound vs I/O-bound operations - Synchronous vs asynchronous bottlenecks - Memory allocation patterns - Database query performance - Network latency impacts - Bundle size and loading performance

Complex scenarios:

Multi-threaded applications
Distributed systems
Microservice latency
Real-time applications
Large dataset processing
Memory leaks over time </think>

Phase 1: Environment Detection

I'll detect the runtime environment and available profiling tools:

#!/bin/bash
# Performance Profiling - Environment Detection

echo "=== Performance Profiling Setup ==="
echo ""

# Create profiling directory
mkdir -p .claude/profiling
PROFILE_DIR=".claude/profiling"
TIMESTAMP=$(date +%Y%m%d-%H%M%S)
REPORT="$PROFILE_DIR/profile-$TIMESTAMP.md"

detect_runtime() {
    local runtime=""

    # Check for Node.js
    if [ -f "package.json" ]; then
        if command -v node >/dev/null 2>&1; then
            runtime="nodejs"
            NODE_VERSION=$(node --version)
            echo "✓ Node.js detected: $NODE_VERSION"
        fi
    fi

    # Check for Python
    if [ -f "requirements.txt" ] || [ -f "setup.py" ] || [ -f "pyproject.toml" ]; then
        if command -v python >/dev/null 2>&1; then
            runtime="python"
            PYTHON_VERSION=$(python --version)
            echo "✓ Python detected: $PYTHON_VERSION"
        elif command -v python3 >/dev/null 2>&1; then
            runtime="python"
            PYTHON_VERSION=$(python3 --version)
            echo "✓ Python detected: $PYTHON_VERSION"
        fi
    fi

    # Check for browser-based app
    if [ -f "package.json" ]; then
        if grep -q "\"react\"\|\"vue\"\|\"angular\"\|\"svelte\"" package.json; then
            if [ -z "$runtime" ]; then
                runtime="browser"
            else
                runtime="$runtime,browser"
            fi
            echo "✓ Frontend framework detected - browser profiling available"
        fi
    fi

    echo "$runtime"
}

RUNTIME=$(detect_runtime)

if [ -z "$RUNTIME" ]; then
    echo "❌ Unable to detect runtime environment"
    echo ""
    echo "Supported environments:"
    echo "  - Node.js applications"
    echo "  - Python applications"
    echo "  - Browser-based applications (React, Vue, Angular, Svelte)"
    exit 1
fi

echo ""
echo "Environment: $RUNTIME"

Phase 2: Install Profiling Tools

I'll check and install necessary profiling tools:

echo ""
echo "=== Installing Profiling Tools ==="

install_profiling_tools() {
    case "$RUNTIME" in
        *nodejs*)
            echo "Setting up Node.js profiling tools..."

            # Check for clinic.js
            if ! npm list -g clinic >/dev/null 2>&1; then
                echo "Installing clinic.js (performance profiling suite)..."
                echo "  npm install -g clinic"
                echo ""
                echo "Clinic.js tools:"
                echo "  - clinic doctor   - Diagnose performance issues"
                echo "  - clinic flame    - CPU flame graphs"
                echo "  - clinic bubbleprof - Async operations"
                echo "  - clinic heapprofiler - Memory profiling"
            else
                echo "✓ clinic.js already installed"
            fi

            # Check for autocannon (benchmarking)
            if ! npm list -g autocannon >/dev/null 2>&1; then
                echo ""
                echo "Installing autocannon (HTTP benchmarking)..."
                echo "  npm install -g autocannon"
            else
                echo "✓ autocannon already installed"
            fi

            echo ""
            echo "Node.js profiling methods available:"
            echo "  1. Built-in --inspect flag"
            echo "  2. clinic.js suite"
            echo "  3. Chrome DevTools"
            ;;

        *python*)
            echo "Setting up Python profiling tools..."

            # Check for profiling packages
            if ! python -c "import cProfile" 2>/dev/null; then
                echo "cProfile not available (should be in stdlib)"
            else
                echo "✓ cProfile available (built-in)"
            fi

            # Check for line_profiler
            if ! python -c "import line_profiler" 2>/dev/null; then
                echo ""
                echo "Install line_profiler for line-by-line profiling:"
                echo "  pip install line_profiler"
            else
                echo "✓ line_profiler available"
            fi

            # Check for memory_profiler
            if ! python -c "import memory_profiler" 2>/dev/null; then
                echo ""
                echo "Install memory_profiler for memory analysis:"
                echo "  pip install memory_profiler"
            else
                echo "✓ memory_profiler available"
            fi

            echo ""
            echo "Python profiling methods available:"
            echo "  1. cProfile (CPU profiling)"
            echo "  2. line_profiler (line-by-line)"
            echo "  3. memory_profiler (memory usage)"
            ;;

        *browser*)
            echo "Browser profiling setup..."
            echo ""
            echo "Browser profiling tools:"
            echo "  1. Chrome DevTools Performance tab"
            echo "  2. Lighthouse CI"
            echo "  3. webpack-bundle-analyzer"
            echo "  4. React DevTools Profiler"

            # Check for bundle analyzer
            if [ -f "package.json" ]; then
                if ! grep -q "webpack-bundle-analyzer" package.json; then
                    echo ""
                    echo "Install webpack-bundle-analyzer:"
                    echo "  npm install --save-dev webpack-bundle-analyzer"
                fi
            fi
            ;;
    esac
}

install_profiling_tools

Phase 3: Node.js Profiling

For Node.js applications, I'll set up and run profiling:

if [[ "$RUNTIME" == *"nodejs"* ]]; then
    echo ""
    echo "=== Node.js Performance Profiling ==="

    # Create profiling scripts
    cat > "$PROFILE_DIR/profile-node.sh" << 'NODEPROFILE'
#!/bin/bash
# Node.js Performance Profiling

echo "Node.js Profiling Options:"
echo ""
echo "1. CPU Profiling (clinic doctor)"
echo "2. Flame Graph (clinic flame)"
echo "3. Async Operations (clinic bubbleprof)"
echo "4. Memory Profiling (clinic heapprofiler)"
echo "5. Built-in V8 Profiler (--prof)"
echo "6. HTTP Load Testing (autocannon)"
echo ""

# Option 1: Clinic Doctor (comprehensive diagnosis)
clinic_doctor() {
    echo "Running clinic doctor..."
    echo "This will start your app and collect performance data"
    echo ""

    # Find entry point
    ENTRY=$(node -p "require('./package.json').main || 'index.js'")

    clinic doctor -- node "$ENTRY"

    echo ""
    echo "✓ Profile complete! Opening report in browser..."
}

# Option 2: Flame Graph
clinic_flame() {
    echo "Running clinic flame..."
    echo "Generating CPU flame graph"
    echo ""

    ENTRY=$(node -p "require('./package.json').main || 'index.js'")

    clinic flame -- node "$ENTRY"

    echo ""
    echo "✓ Flame graph generated!"
}

# Option 3: Async Operations
clinic_bubble() {
    echo "Running clinic bubbleprof..."
    echo "Analyzing async operations"
    echo ""

    ENTRY=$(node -p "require('./package.json').main || 'index.js'")

    clinic bubbleprof -- node "$ENTRY"

    echo ""
    echo "✓ Async analysis complete!"
}

# Option 4: V8 Profiler
v8_profiler() {
    echo "Running V8 profiler..."
    echo ""

    ENTRY=$(node -p "require('./package.json').main || 'index.js'")

    node --prof "$ENTRY"

    # Process the profile
    PROFILE=$(ls isolate-*.log | head -1)
    if [ -n "$PROFILE" ]; then
        node --prof-process "$PROFILE" > processed-profile.txt
        echo "✓ Profile processed: processed-profile.txt"
    fi
}

# Option 5: HTTP Load Test
http_load_test() {
    echo "HTTP Load Testing with autocannon"
    echo ""
    read -p "Enter URL to test (e.g., http://localhost:3000): " URL
    read -p "Duration in seconds (default: 10): " DURATION
    DURATION=${DURATION:-10}

    echo ""
    echo "Running load test..."
    autocannon -d $DURATION "$URL"
}

# Main menu
case "${1:-1}" in
    1) clinic_doctor ;;
    2) clinic_flame ;;
    3) clinic_bubble ;;
    4) clinic_heapprofiler -- node $(node -p "require('./package.json').main || 'index.js'") ;;
    5) v8_profiler ;;
    6) http_load_test ;;
    *) echo "Invalid option" ;;
esac
NODEPROFILE

    chmod +x "$PROFILE_DIR/profile-node.sh"
    echo "✓ Created Node.js profiling script: $PROFILE_DIR/profile-node.sh"

    # Identify potential bottlenecks in code
    echo ""
    echo "Scanning for potential performance issues..."

    # Check for synchronous blocking operations
    SYNC_BLOCKING=$(grep -r "readFileSync\|writeFileSync\|execSync" \
        --include="*.js" --include="*.ts" \
        --exclude-dir=node_modules \
        --exclude-dir=dist \
        . 2>/dev/null | wc -l)

    if [ "$SYNC_BLOCKING" -gt 0 ]; then
        echo "⚠️  Found $SYNC_BLOCKING synchronous blocking operations"
        echo "   Consider using async versions (readFile, writeFile, exec)"
    fi

    # Check for missing await
    MISSING_AWAIT=$(grep -r "^\s*async.*function" \
        --include="*.js" --include="*.ts" \
        --exclude-dir=node_modules \
        . 2>/dev/null | wc -l)

    echo "💡 Async functions found: $MISSING_AWAIT"
    echo "   Verify all async calls use await or proper promise handling"

    # Check for large dependencies
    echo ""
    echo "Analyzing bundle size..."
    if [ -f "package.json" ]; then
        DEPS_COUNT=$(cat package.json | jq '.dependencies | length' 2>/dev/null || echo "N/A")
        echo "  Dependencies: $DEPS_COUNT"

        # Suggest bundle analyzer
        echo ""
        echo "To analyze bundle size:"
        echo "  npm install --save-dev webpack-bundle-analyzer"
        echo "  # Add to webpack config and run build"
    fi
fi

Phase 4: Python Profiling

For Python applications, I'll set up profiling:

if [[ "$RUNTIME" == *"python"* ]]; then
    echo ""
    echo "=== Python Performance Profiling ==="

    cat > "$PROFILE_DIR/profile-python.sh" << 'PYTHONPROFILE'
#!/bin/bash
# Python Performance Profiling

echo "Python Profiling Options:"
echo ""
echo "1. cProfile (CPU profiling)"
echo "2. line_profiler (line-by-line)"
echo "3. memory_profiler (memory usage)"
echo ""

# Option 1: cProfile
cpu_profile() {
    echo "Running cProfile..."
    echo ""

    # Find main Python file
    if [ -f "main.py" ]; then
        ENTRY="main.py"
    elif [ -f "app.py" ]; then
        ENTRY="app.py"
    else
        read -p "Enter Python file to profile: " ENTRY
    fi

    python -m cProfile -o profile.stats "$ENTRY"

    # Analyze profile
    python << 'ANALYZE'
import pstats
from pstats import SortKey

p = pstats.Stats('profile.stats')
print("\n=== Top 20 Functions by Cumulative Time ===")
p.sort_stats(SortKey.CUMULATIVE).print_stats(20)

print("\n=== Top 20 Functions by Total Time ===")
p.sort_stats(SortKey.TIME).print_stats(20)
ANALYZE

    echo ""
    echo "✓ Profile saved to: profile.stats"
}

# Option 2: line_profiler
line_profile() {
    echo "Using line_profiler..."
    echo ""
    echo "Add @profile decorator to functions you want to profile"
    echo "Example:"
    echo "  @profile"
    echo "  def slow_function():"
    echo "      pass"
    echo ""

    read -p "Enter Python file with @profile decorators: " ENTRY

    kernprof -l -v "$ENTRY"
}

# Option 3: memory_profiler
memory_profile() {
    echo "Using memory_profiler..."
    echo ""
    echo "Add @profile decorator to functions you want to profile"
    echo ""

    read -p "Enter Python file with @profile decorators: " ENTRY

    python -m memory_profiler "$ENTRY"
}

case "${1:-1}" in
    1) cpu_profile ;;
    2) line_profile ;;
    3) memory_profile ;;
    *) echo "Invalid option" ;;
esac
PYTHONPROFILE

    chmod +x "$PROFILE_DIR/profile-python.sh"
    echo "✓ Created Python profiling script: $PROFILE_DIR/profile-python.sh"

    # Identify potential bottlenecks
    echo ""
    echo "Scanning for potential performance issues..."

    # Check for list comprehensions that could be generators
    LIST_COMP=$(grep -r "\[.*for .* in .*\]" \
        --include="*.py" \
        --exclude-dir=venv \
        --exclude-dir=env \
        . 2>/dev/null | wc -l)

    echo "💡 List comprehensions found: $LIST_COMP"
    echo "   Consider using generators for large datasets"

    # Check for database queries without indexing hints
    DB_QUERIES=$(grep -r "SELECT.*FROM\|\.filter(\|\.get(" \
        --include="*.py" \
        . 2>/dev/null | wc -l)

    if [ "$DB_QUERIES" -gt 0 ]; then
        echo "💡 Database queries found: $DB_QUERIES"
        echo "   Ensure proper indexing and use select_related/prefetch_related"
    fi
fi

Phase 5: Browser Profiling

For frontend applications, I'll set up browser profiling:

if [[ "$RUNTIME" == *"browser"* ]]; then
    echo ""
    echo "=== Browser Performance Profiling ==="

    cat > "$PROFILE_DIR/profile-browser.md" << 'BROWSERPROFILE'
# Browser Performance Profiling Guide

## Chrome DevTools Performance Profiling

1. **Open Chrome DevTools**
   - F12 or Right-click → Inspect
   - Go to "Performance" tab

2. **Record Performance**
   - Click Record button (circle)
   - Interact with your app (navigate, click, scroll)
   - Click Stop

3. **Analyze Results**
   - Look for long tasks (> 50ms)
   - Identify JavaScript execution time
   - Check layout/reflow operations
   - Analyze network waterfall

4. **Key Metrics**
   - FCP (First Contentful Paint) - < 1.8s
   - LCP (Largest Contentful Paint) - < 2.5s
   - TBT (Total Blocking Time) - < 200ms
   - CLS (Cumulative Layout Shift) - < 0.1

## Lighthouse Audit

```bash
# Install Lighthouse CLI
npm install -g lighthouse

# Run audit
lighthouse http://localhost:3000 --view

# Or programmatically
npx lighthouse http://localhost:3000 --output html --output-path ./lighthouse-report.html

Bundle Analysis

Webpack Bundle Analyzer

npm install --save-dev webpack-bundle-analyzer

# Add to webpack.config.js
const BundleAnalyzerPlugin = require('webpack-bundle-analyzer').BundleAnalyzerPlugin;

module.exports = {
  plugins: [
    new BundleAnalyzerPlugin()
  ]
}

# Run build
npm run build

Vite Bundle Analysis

npm install --save-dev rollup-plugin-visualizer

# Add to vite.config.js
import { visualizer } from 'rollup-plugin-visualizer';

export default {
  plugins: [
    visualizer({ open: true })
  ]
}

# Run build
npm run build

React Profiler

import { Profiler } from 'react';

function onRenderCallback(
  id, phase, actualDuration, baseDuration, startTime, commitTime
) {
  console.log(`${id} took ${actualDuration}ms to render`);
}

<Profiler id="MyComponent" onRender={onRenderCallback}>
  <MyComponent />
</Profiler>

Common Performance Issues

1. Unnecessary Re-renders

Use React.memo() for expensive components
Implement shouldComponentUpdate or use useMemo
Avoid inline function definitions in JSX

2. Large Bundle Size

Code splitting with React.lazy()
Tree shaking unused code
Analyze and remove large dependencies

3. Unoptimized Images

Use WebP format
Implement lazy loading
Add proper width/height attributes

4. Too Many Network Requests

Bundle similar resources
Use HTTP/2 multiplexing
Implement resource hints (preload, prefetch)

5. Blocking JavaScript

Defer non-critical scripts
Use async attribute
Move scripts to bottom of body

Optimization Checklist

Bundle size < 200KB (gzipped)
First load < 3 seconds
Images optimized and lazy-loaded
Code split by route
Critical CSS inlined
Font loading optimized
Service worker for caching
Compression enabled (gzip/brotli) BROWSERPROFILE

echo "✓ Created browser profiling guide: $PROFILE_DIR/profile-browser.md"

Check for performance anti-patterns

echo "" echo "Scanning for performance anti-patterns..."

Check for console.log in production

CONSOLE_LOGS=$(grep -r "console.log|console.debug"
--include=".js" --include=".jsx" --include=".ts" --include=".tsx"
--exclude-dir=node_modules
. 2>/dev/null | wc -l)

if [ "$CONSOLE_LOGS" -gt 10 ]; then echo "⚠️ Found $CONSOLE_LOGS console.log statements" echo " Remove or use conditional logging for production" fi

Check for large images

if [ -d "public" ] || [ -d "static" ] || [ -d "assets" ]; then echo "" echo "Checking for large images..." find public static assets -type f ( -name ".jpg" -o -name ".png" -o -name "*.jpeg" ) -size +500k 2>/dev/null | while read img; do SIZE=$(du -h "$img" | cut -f1) echo " ⚠️ Large image: $img ($SIZE)" done fi fi


## Phase 6: Generate Performance Report

I'll create a comprehensive performance analysis report:

```bash
echo ""
echo "=== Generating Performance Report ==="

cat > "$REPORT" << EOF
# Performance Profile Report

**Generated:** $(date)
**Runtime:** $RUNTIME
**Project:** $(basename $(pwd))

---

## Profiling Setup

### Environment

- **Runtime:** $RUNTIME
- **Node Version:** ${NODE_VERSION:-N/A}
- **Python Version:** ${PYTHON_VERSION:-N/A}

### Tools Available

EOF

# Add tool-specific sections
if [[ "$RUNTIME" == *"nodejs"* ]]; then
    cat >> "$REPORT" << 'EOF'

#### Node.js Profiling Tools

- clinic.js suite (doctor, flame, bubbleprof, heapprofiler)
- V8 built-in profiler (--prof)
- Chrome DevTools inspector
- autocannon (HTTP benchmarking)

**Run profiling:**
```bash
./claude/profiling/profile-node.sh 1  # Clinic doctor
./claude/profiling/profile-node.sh 2  # Flame graph
./claude/profiling/profile-node.sh 6  # Load test

EOF fi

if [[ "$RUNTIME" == "python" ]]; then cat >> "$REPORT" << 'EOF'

Python Profiling Tools

cProfile (CPU profiling)
line_profiler (line-by-line analysis)
memory_profiler (memory usage)

Run profiling:

./claude/profiling/profile-python.sh 1  # cProfile
./claude/profiling/profile-python.sh 2  # line_profiler

EOF fi

if [[ "$RUNTIME" == "browser" ]]; then cat >> "$REPORT" << 'EOF'

Browser Profiling Tools

Chrome DevTools Performance
Lighthouse CI
webpack-bundle-analyzer
React DevTools Profiler

See guide:

cat .claude/profiling/profile-browser.md

EOF fi

cat >> "$REPORT" << 'EOF'

Quick Wins

Immediate Optimizations

Enable Compression
- gzip/brotli compression
- Reduce bundle size by 60-80%
Code Splitting
- Split by route
- Lazy load heavy components
- Reduce initial load time
Optimize Images
- Use WebP format
- Lazy loading
- Proper sizing
Remove Unnecessary Dependencies
- Audit package.json
- Replace heavy libraries with lighter alternatives
- Tree shake unused code
Database Query Optimization
- Add proper indexes
- Use connection pooling
- Implement caching

Performance Monitoring

Continuous Monitoring

Set up Lighthouse CI in your pipeline
Monitor Core Web Vitals
Track bundle size over time
Profile in production

Recommended Tools

Next Steps

Resources

EOF

echo "✓ Created performance report: $REPORT"


## Summary

```bash
echo ""
echo "=== ✓ Performance Profiling Setup Complete ==="
echo ""
echo "🎯 Runtime Environment: $RUNTIME"
echo ""
echo "📁 Generated files:"
echo "  - $REPORT"

if [[ "$RUNTIME" == *"nodejs"* ]]; then
    echo "  - $PROFILE_DIR/profile-node.sh"
fi

if [[ "$RUNTIME" == *"python"* ]]; then
    echo "  - $PROFILE_DIR/profile-python.sh"
fi

if [[ "$RUNTIME" == *"browser"* ]]; then
    echo "  - $PROFILE_DIR/profile-browser.md"
fi

echo ""
echo "🚀 Quick Start:"
echo ""

if [[ "$RUNTIME" == *"nodejs"* ]]; then
    echo "Node.js Profiling:"
    echo "  $PROFILE_DIR/profile-node.sh 1  # Comprehensive diagnosis"
    echo "  $PROFILE_DIR/profile-node.sh 2  # CPU flame graph"
    echo "  $PROFILE_DIR/profile-node.sh 6  # HTTP load test"
    echo ""
fi

if [[ "$RUNTIME" == *"python"* ]]; then
    echo "Python Profiling:"
    echo "  $PROFILE_DIR/profile-python.sh 1  # CPU profiling"
    echo "  $PROFILE_DIR/profile-python.sh 2  # Line-by-line"
    echo ""
fi

if [[ "$RUNTIME" == *"browser"* ]]; then
    echo "Browser Profiling:"
    echo "  lighthouse http://localhost:3000 --view"
    echo "  cat $PROFILE_DIR/profile-browser.md  # Full guide"
    echo ""
fi

echo "📊 Performance Analysis Workflow:"
echo ""
echo "1. Establish baseline performance"
echo "   - Run profiling on current code"
echo "   - Record key metrics (response time, memory, CPU)"
echo ""
echo "2. Identify bottlenecks"
echo "   - Find slowest functions/routes"
echo "   - Check database queries"
echo "   - Analyze bundle size"
echo ""
echo "3. Optimize iteratively"
echo "   - Fix highest-impact issues first"
echo "   - Re-profile after each change"
echo "   - Measure improvement"
echo ""
echo "4. Set performance budgets"
echo "   - Bundle size limits"
echo "   - Response time targets"
echo "   - Memory usage constraints"
echo ""
echo "💡 Common Bottlenecks:"
echo "  - Synchronous blocking operations"
echo "  - N+1 database queries"
echo "  - Large bundle sizes"
echo "  - Unoptimized images"
echo "  - Missing caching"
echo "  - Inefficient algorithms"
echo ""
echo "🔗 Integration Points:"
echo "  - /debug-root-cause - Investigate performance issues"
echo "  - /test - Add performance tests"
echo "  - /ci-setup - Add performance budgets to CI"
echo ""
echo "View full report: cat $REPORT"

Best Practices

Profiling Strategy:

Profile in production-like environment
Use representative workloads
Run multiple samples for accuracy
Focus on 80/20 rule (biggest bottlenecks first)
Measure before and after optimizations

Common Optimizations:

Cache frequently accessed data
Use connection pooling for databases
Implement lazy loading
Optimize database queries with indexes
Use CDN for static assets
Enable compression
Minimize bundle size

Performance Budgets:

Set and enforce limits on bundle size
Target response times for critical paths
Monitor memory usage patterns
Track Core Web Vitals (LCP, FID, CLS)

Credits: Performance profiling methodology based on Node.js profiling guide, Python performance documentation, Chrome DevTools documentation, and Web.dev performance best practices. Clinic.js integration patterns from NearForm. Lighthouse CI guidance from Google Chrome team.

performance-profile

Safety Notice

Copy this and send it to your AI assistant to learn

Performance Profiling & Bottleneck Detection

Token Optimization

Optimization Strategies

1. Runtime Detection Caching (Saves 700 tokens per invocation)

2. Early Exit for Configured Environments (Saves 85%)

3. Script Generation on Demand (Saves 70%)

4. Sample-Based Anti-Pattern Detection (Saves 80%)

5. Grep-Based Pattern Detection (Saves 90%)

6. Bash-Based Tool Execution (Saves 70%)

7. Minimal Report Generation (Saves 60%)

8. Tool Installation Guidance Only (Saves 95%)

Cache Invalidation

Real-World Token Usage

Progressive Disclosure

Implementation Notes

Extended Thinking for Performance Analysis

Phase 1: Environment Detection

Phase 2: Install Profiling Tools

Phase 3: Node.js Profiling

Phase 4: Python Profiling

Phase 5: Browser Profiling

Bundle Analysis

Webpack Bundle Analyzer

Vite Bundle Analysis

React Profiler

Common Performance Issues

1. Unnecessary Re-renders

2. Large Bundle Size

3. Unoptimized Images

4. Too Many Network Requests

5. Blocking JavaScript

Optimization Checklist

Check for performance anti-patterns

Check for console.log in production

Check for large images

Python Profiling Tools

Browser Profiling Tools

Quick Wins

Immediate Optimizations

Performance Monitoring

Continuous Monitoring

Recommended Tools

Next Steps

Resources

Best Practices

Source Transparency

Related Skills

sessions-init

cache-strategy

postman-convert