agentsclimarketplace

Brightdata cost tuning

Skill jeremylongshore/claude-code-plugins-plus-skills/plugins/saas-packs/brightdata-pack/skills/brightdata-cost-tuning

'Optimize Bright Data costs through tier selection, sampling, and usage monitoring.From its SKILL.md

Install
npx -y skills add jeremylongshore/claude-code-plugins-plus-skills --skill brightdata-cost-tuning

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

What its file declares

Copied from the file, not written here

The file declares its own license as MIT. That is the author’s claim about this one file, and it is not the same thing as the license GitHub reports for the repository, which is listed with the other numbers below.

SKILL.md

6.4 KB, ~1.5k tokens by cl100k_base, as published. Nobody here has run it

Bright Data Cost Tuning

Overview

Optimize Bright Data costs through product selection, caching, and usage monitoring. Bright Data charges per request (Web Unlocker, SERP API), per GB (Residential Proxy), or per page (Datasets). Choosing the right product and avoiding redundant requests is the primary cost lever.

Prerequisites

  • Access to Bright Data billing dashboard
  • Understanding of current scraping volumes
  • Usage monitoring configured (optional)

Pricing Model

ProductPricingTypical CostBest For
Residential ProxyPer GB transferred$8-15/GBHigh-volume, simple pages
Web UnlockerPer successful request$1-3/1000 reqAnti-bot protected sites
Scraping BrowserPer browser session$5-10/1000 sessionsJS-heavy SPAs
SERP APIPer search$2-5/1000 searchesSearch engine results
Datasets (pre-built)Per record$0.001-0.01/recordBulk data (Amazon, LinkedIn)
Web Scraper APIPer pageVaries by datasetCustom async scraping

Instructions

Step 1: Product Selection Cost Matrix

function estimateMonthlyCost(config: {
  product: 'residential' | 'web_unlocker' | 'scraping_browser' | 'serp_api';
  requestsPerMonth: number;
  avgPageSizeKB?: number;
}) {
  switch (config.product) {
    case 'residential':
      const gbTransferred = (config.requestsPerMonth * (config.avgPageSizeKB || 200)) / 1_000_000;
      return { cost: gbTransferred * 10, unit: 'GB', quantity: gbTransferred };
    case 'web_unlocker':
      return { cost: config.requestsPerMonth * 0.002, unit: 'requests', quantity: config.requestsPerMonth };
    case 'scraping_browser':
      return { cost: config.requestsPerMonth * 0.008, unit: 'sessions', quantity: config.requestsPerMonth };
    case 'serp_api':
      return { cost: config.requestsPerMonth * 0.003, unit: 'searches', quantity: config.requestsPerMonth };
  }
}

// Example: 50,000 product pages/month
console.log(estimateMonthlyCost({ product: 'web_unlocker', requestsPerMonth: 50000 }));
// { cost: 100, unit: 'requests', quantity: 50000 }
console.log(estimateMonthlyCost({ product: 'residential', requestsPerMonth: 50000, avgPageSizeKB: 300 }));
// { cost: 150, unit: 'GB', quantity: 15 }

Step 2: Reduce Costs with Caching

// Response caching is the single biggest cost saver
// Cache policy by data freshness requirements
const CACHE_TTLS = {
  product_price: 3600000,     // 1 hour — prices change frequently
  product_details: 86400000,  // 24 hours — descriptions rarely change
  search_results: 1800000,    // 30 minutes — SERPs change often
  static_page: 604800000,     // 7 days — about/contact pages
};

// Track cache savings
let cacheSavings = 0;
function trackCacheHit(product: string) {
  const costPerRequest = { web_unlocker: 0.002, scraping_browser: 0.008, serp_api: 0.003 };
  cacheSavings += costPerRequest[product] || 0.002;
  console.log(`Cache savings this session: $${cacheSavings.toFixed(4)}`);
}

Step 3: Use Bulk APIs for Volume Jobs

// Individual requests: 50,000 requests * $0.002 = $100
// Web Scraper API: 1 trigger with 50,000 URLs = typically cheaper (volume discounts)

async function bulkScrapeForCost(urls: string[]) {
  // Batch into single trigger — one API call, lower cost
  const response = await fetch(
    `https://api.brightdata.com/datasets/v3/trigger?dataset_id=${DATASET_ID}&format=json`,
    {
      method: 'POST',
      headers: {
        'Authorization': `Bearer ${process.env.BRIGHTDATA_API_TOKEN}`,
        'Content-Type': 'application/json',
      },
      body: JSON.stringify(urls.map(url => ({ url }))),
    }
  );
  return response.json();
}

Step 4: Usage Monitoring

class BrightDataUsageTracker {
  private dailyRequests = 0;
  private dailyCost = 0;
  private readonly budgetAlert: number;

  constructor(dailyBudgetUSD: number) {
    this.budgetAlert = dailyBudgetUSD * 0.8;
  }

  track(product: string) {
    this.dailyRequests++;
    const costs = { web_unlocker: 0.002, scraping_browser: 0.008, serp_api: 0.003, residential: 0.0001 };
    this.dailyCost += costs[product] || 0.002;

    if (this.dailyCost > this.budgetAlert) {
      console.warn(`BUDGET ALERT: Daily cost $${this.dailyCost.toFixed(2)} exceeds 80% of budget`);
    }
  }

  report() {
    return {
      requests: this.dailyRequests,
      estimatedCost: `$${this.dailyCost.toFixed(2)}`,
      projectedMonthly: `$${(this.dailyCost * 30).toFixed(2)}`,
    };
  }
}

Step 5: Cost Reduction Checklist

  • Cache responses to avoid re-scraping same URLs
  • Use Residential Proxy for simple pages (cheaper per request)
  • Use Web Scraper API for 100+ URL bulk jobs
  • Use Datasets API for common targets (Amazon, LinkedIn) — pre-built scrapers
  • Set budget alerts in Bright Data CP > Billing
  • Monitor daily usage with tracker class above
  • Avoid Scraping Browser for pages that don't need JavaScript

Output

  • Product selection matching cost requirements
  • Response caching reducing redundant requests
  • Budget monitoring and alerting
  • Projected monthly cost estimates

Error Handling

IssueCauseSolution
Unexpected chargesUsing expensive product for simple pagesSwitch to Residential Proxy
Budget exceededNo monitoringImplement usage tracker
Overpaying for dataScraping what Datasets API providesCheck pre-built datasets first
High per-request costNo cachingAdd response cache (biggest lever)

Resources

Next Steps

For architecture patterns, see brightdata-reference-architecture.

What ships with it

Read from the repository

Just SKILL.md. No reference files, no scripts.

Keep looking

Skills are one crate of 326,144. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.