scraperapi-agent-onboarding

ScraperAPI — Agent Onboarding

ScraperAPI gives agents and apps reliable access to web data: any URL returned as clean HTML or markdown (with proxy rotation, CAPTCHA bypass, and JS rendering handled automatically), structured JSON from 20+ supported platforms (Amazon, Walmart, eBay, Redfin, Google SERP), and a crawler for multi-page extraction.

This skill is the entry point. Read it once, pick a path, then hand off to the narrower skill that owns that path.

Getting an API key

All paths require a ScraperAPI key. If you don't have one:

Sign up at https://www.scraperapi.com/ — free trial includes 5,000 credits
Copy your key from https://dashboard.scraperapi.com/
Set it in your environment — never hardcode it in code or configs:

# macOS / Linux
export SCRAPERAPI_API_KEY=<your-key>

# Windows PowerShell
$env:SCRAPERAPI_API_KEY = "<your-key>"

# .env file
SCRAPERAPI_API_KEY=...

Verify the key works before doing real work:

curl "https://api.scraperapi.com/?api_key=$SCRAPERAPI_API_KEY&url=https://httpbin.org/ip"

A JSON response with an origin IP confirms the key is valid. A 401 means the key is wrong or inactive — do not retry with the same key, surface the error.

Choose your path

Situation	Path
LLM agent needs web data during this session	Path A — MCP Server
Adding ScraperAPI to app code	Path B — SDK / REST integration
Quick request, or environment without Node/Python	Path C — REST API directly
Need an API key first	Path D — Auth only (above)

If your task spans paths, do them in order: get key → MCP or SDK → test one real request before building.

Path A — MCP Server (LLM tool layer)

Use this when the consumer is an LLM agent (Claude Code, custom agent loop) that should call ScraperAPI as tools. The MCP server exposes up to 22 tools — scrape, Google, Amazon, Walmart, eBay, Redfin, and crawler — over a single connection.

Remote server (recommended)

No installation required. All 22 tools available.

Add to your claude_desktop_config.json, Claude Code project settings, or MCP client config:

{
  "mcpServers": {
    "ScraperAPI": {
      "command": "npx",
      "args": [
        "mcp-remote",
        "https://mcp.scraperapi.com/mcp",
        "--header",
        "Authorization: Bearer ${SCRAPERAPI_API_KEY}"
      ]
    }
  }
}

Requires SCRAPERAPI_API_KEY set in the environment. npx mcp-remote handles the remote connection — no separate install step needed.

Local server (self-hosted, scrape-only)

If you prefer a local server or the remote endpoint is unavailable. Requires Python 3.11+.

pip install scraperapi-mcp-server

{
  "mcpServers": {
    "ScraperAPI": {
      "command": "python",
      "args": ["-m", "scraperapi_mcp_server"],
      "env": { "API_KEY": "<YOUR_SCRAPERAPI_API_KEY>" }
    }
  }
}

The local server exposes only the scrape tool. For structured Google search data on the local variant, use scrape with autoparse: true and outputFormat: "json" on a Google search URL.

After MCP setup

Hand off to the scraperapi-mcp skill. It owns tool selection, parameter optimization, credit cost guidance, and error recovery for all 22 MCP tools. Also check the MCP setup reference for variant detection and troubleshooting.

Path B — SDK / REST integration (app code)

Use this when you're building an application, script, or workflow that calls ScraperAPI from code.

Smoke test first

Always run one real request before writing integration logic — catches auth, credit, and parameter issues before they hide inside your app's error paths:

import os, requests

r = requests.get(
    "https://api.scraperapi.com/",
    params={"api_key": os.environ["SCRAPERAPI_API_KEY"], "url": "https://httpbin.org/ip"}
)
print(r.status_code, r.text[:200])

A 200 with valid JSON confirms you're wired up.

Pick the right API

Job in the app	API	Key docs
Fetch any URL as HTML or markdown	Standard API	docs
Render JS-heavy pages (React, Vue, SPAs)	Standard API + `render=true`	docs
Structured JSON: Amazon, Google, Walmart, eBay, Redfin	Structured Data API	docs
Batch scraping / 100+ URLs	Async Jobs API	docs
Recurring, scheduled scraping without managing infra	DataPipeline (dashboard)	docs

Python

pip install scraperapi-sdk

import os
from scraperapi_sdk import ScraperAPIClient

client = ScraperAPIClient(os.environ["SCRAPERAPI_API_KEY"])
result = client.get("https://example.com", render=True, country_code="us")
print(result)

Node.js

npm install scraperapi-sdk

const scraperapi = require('scraperapi-sdk')(process.env.SCRAPERAPI_API_KEY);
const result = await scraperapi.get('https://example.com', { render: true, country_code: 'us' });
console.log(result);

PHP, Ruby, and Java SDKs are also available — see docs.scraperapi.com for setup guides.

Path C — REST API (no SDK required)

Use this when the environment can't run pip or npm, or when you only need a few requests.

Base URL: https://api.scraperapi.com/ Auth: api_key query parameter (or apiKey in the JSON body for async)

# Basic scrape
curl "https://api.scraperapi.com/?api_key=$SCRAPERAPI_API_KEY&url=https://example.com"

# With JS rendering
curl "https://api.scraperapi.com/?api_key=$SCRAPERAPI_API_KEY&url=https://example.com&render=true"

# Structured data — Google SERP
curl "https://api.scraperapi.com/structured/google/search?api_key=$SCRAPERAPI_API_KEY&query=web+scraping"

# Structured data — Amazon product
curl "https://api.scraperapi.com/structured/amazon/product?api_key=$SCRAPERAPI_API_KEY&asin=B09V3KXJPB"

# Submit an async job
curl -X POST "https://async.scraperapi.com/jobs" \
  -H "Content-Type: application/json" \
  -d "{\"apiKey\": \"$SCRAPERAPI_API_KEY\", \"url\": \"https://example.com\"}"

# Poll async job status
curl "https://async.scraperapi.com/jobs/<jobId>?apiKey=$SCRAPERAPI_API_KEY"

For the full parameter surface (premium proxies, country targeting, session stickiness, binary targets), see the API reference.

Common status codes

Code	Meaning	Action
`200`	Success	Return / parse the body
`401`	Invalid API key	Surface error — do not retry
`403`	Blocked by target site	Retry with `premium=true`, `ultra_premium=true`, or a different `country_code`
`429`	Concurrent limit hit	Back off; switch to async for batch work
`500`/`503`	Transient failure	Retry with exponential backoff (3–5 attempts)

After onboarding — where to go next

Route to the narrowest skill or reference that fits the task:

User says…	Go to
"scrape this URL" / "get this page"	`scraperapi-mcp` → `scrape` tool
"search Google for…" / "find URLs about…"	`scraperapi-mcp` → `google_search`
"get Amazon / Walmart / eBay data"	`scraperapi-mcp` → SDE tools
"real estate / Redfin listings"	`scraperapi-mcp` → Redfin tools
"crawl a website / extract all links"	`scraperapi-mcp` → `crawler_job_start`
"scrape 100+ URLs / batch job"	Async Jobs API — Path C above
"schedule recurring scraping"	DataPipeline — https://docs.scraperapi.com/
"debug a failed scrape or blocked request"	`scraperapi-mcp` error recovery section

When in doubt, prefer structured data endpoints over raw HTML scraping for supported platforms — they return clean JSON with no parsing logic, are more reliable, and often cost fewer credits than raw scraping with render=true.