Skip to content

SETUP.md

Latest commit

 

History

History
323 lines (241 loc) · 5.54 KB

SETUP.md

File metadata and controls

323 lines (241 loc) · 5.54 KB

Setup & Installation Guide

Quick Start (5 minutes)

1. Install

git clone https://github.com/LvcidPsyche/claude-api-optimization.git
cd claude-api-optimization
npm install

2. Set Environment Variables

export ANTHROPIC_API_KEY="your-key-here"
export MODEL_PRIMARY="anthropic/claude-haiku-4-5"
export MODEL_FALLBACK="anthropic/claude-sonnet-4-20250514"

3. Verify Installation

npm test
node demo.js

Full Installation

Prerequisites

  • Node.js ≥ 16.0.0 (install)
  • npm ≥ 8.0.0 (comes with Node.js)
  • Claude API Key (get one)

Step-by-Step

  1. Clone Repository

    git clone https://github.com/LvcidPsyche/claude-api-optimization.git
cd claude-api-optimization

  2. Install Dependencies

    npm install

  3. Configure Environment

    Option A: Environment File

    cp .env.example .env
# Edit .env with your API key

    Option B: Environment Variables

    export ANTHROPIC_API_KEY="sk-ant-..."

  4. Verify Setup

    npm test

Usage

Using Individual Tools

Cost Monitor

const ClaudeCostMonitor = require('./cost-monitor');
const monitor = new ClaudeCostMonitor();

monitor.trackUsage('haiku-4-5', 1000, 500);
console.log(monitor.generateReport());

Model Router

const ModelRouter = require('./model-router');
const router = new ModelRouter();

const result = router.selectModel('Classify this email');
console.log(result.model); // "anthropic/claude-haiku-4-5"
console.log(result.estimatedCostSavings);

Prompt Cache

const PromptCache = require('./prompt-cache');
const cache = new PromptCache();

const optimized = cache.optimizeForCaching(messages, systemPrompt);
console.log(optimized);

Response Cache

const ResponseCache = require('./response-cache');
const cache = new ResponseCache();

cache.set('Query', response);
const cached = cache.get('Query');
console.log(cache.getStats());

Batch Processor

const BatchProcessor = require('./batch-processor');
const processor = new BatchProcessor(process.env.ANTHROPIC_API_KEY);

processor.addRequest('Classify this email');
console.log(processor.estimateMetrics());

Benchmark

const CostBenchmark = require('./benchmark');
const benchmark = new CostBenchmark();

benchmark.runScenario('My scenario', requests);
console.log(benchmark.getReport());

Run Examples

# Email classification
node examples/email-classification.js

# Batch content generation
node examples/batch-content-generation.js

Run Tests

# All tests
npm test

# Specific test file
npm test test/cost-monitor.test.js

# With coverage
npm test -- --coverage

Integration Guide

OpenClaw Integration

# Copy config to OpenClaw
cp optimized-config.json ~/.openclaw/openclaw.json

Node.js Application

const {
  ClaudeCostMonitor,
  ModelRouter,
  PromptCache,
  ResponseCache,
  BatchProcessor,
  CostBenchmark
} = require('claude-api-optimization');

// Use in your app

CLI Usage

# Run cost monitoring
node cost-monitor.js

# Run model routing
node model-router.js

# Run benchmarks
node benchmark.js

Configuration

Environment Variables

ANTHROPIC_API_KEY=your_key_here
MODEL_PRIMARY=anthropic/claude-haiku-4-5
MODEL_FALLBACK=anthropic/claude-sonnet-4-5
CACHE_TTL=3600000
CACHE_MAX_SIZE=1000
BATCH_MAX_SIZE=100000

Config File (optional)

Create config.json:

{
  "models": {
    "primary": "anthropic/claude-haiku-4-5",
    "fallback": "anthropic/claude-sonnet-4-5"
  },
  "cache": {
    "ttl": 3600000,
    "maxSize": 1000
  },
  "batch": {
    "maxSize": 100000,
    "maxWaitTime": 86400000
  }
}

Troubleshooting

Issue: "Missing API Key"

Solution:

export ANTHROPIC_API_KEY="your-key-here"

Issue: "Tests failing"

Solution:

npm install
npm test

Issue: "Module not found"

Solution:

rm -rf node_modules package-lock.json
npm install

Issue: "Cache not working"

Solution:

  • Verify cache TTL is set correctly
  • Check cache size limits
  • Ensure entries haven't expired

Issue: "Batch API timeout"

Solution:

  • Batch processing can take 5-24 hours
  • Check batch status via API
  • Reduce batch size if needed

Performance Tuning

Optimize for Your Use Case

High-Volume Classification:

export CACHE_MAX_SIZE=5000  # Larger cache
export CACHE_TTL=86400000   # 24-hour TTL

Batch Processing:

export BATCH_MAX_SIZE=100000  # Max batch size

Real-time Applications:

export CACHE_TTL=300000      # 5-minute TTL
export CACHE_MAX_SIZE=500    # Smaller cache

Monitoring

View Metrics

const monitor = new ClaudeCostMonitor();
const stats = monitor.generateReport();
console.log(JSON.stringify(stats, null, 2));

Export Metrics

const cache = new ResponseCache();
const metrics = cache.export();
// Send to monitoring system

Support

Next Steps

  1. ✅ Complete installation
  2. 📖 Read README.md
  3. 🧪 Run tests: npm test
  4. 📚 Review examples in examples/
  5. 🚀 Integrate into your application