Self-hosted & MIT
The core engine is MIT open source. Deploy with one command, own your stack end-to-end, and keep your data on your own hardware — no vendor lock-in, no usage tax.
- docker compose up
- your infra, your data
- 100+ files MIT
MIT open-source web scraping and browser automation with a 10-tier Cloudflare bypass, stealth browsing, and x402 pay-per-call. Built for data teams and AI agents — no API keys, no credit cards, no SaaS lock-in.
15 data domains42 MCP toolsmulti-chain x402
runs on :8005 · FlareSolverr on :8192 · one command to local
{
"url": "https://store.example/product",
"title": "AeroLite Runner — $129",
"bypass": "flaresolverr",
"status": 200,
"took_ms": 1843,
"markdown": "# AeroLite Runner\n\nUltra-light..."
}Why Pry
The core engine is MIT open source. Deploy with one command, own your stack end-to-end, and keep your data on your own hardware — no vendor lock-in, no usage tax.
Camoufox, TLS-fingerprint randomization, and six behavioral JavaScript scripts make automated traffic look human. Escalates from cheap direct requests to a full stealth browser automatically.
Gate any endpoint behind HTTP 402. Agents and users pay per call in USDC, USDT, DAI or native assets across 13 chains — settled on-chain. No subscriptions, no Stripe, no KYC.
Capabilities
Direct HTTP → TLS fingerprint → cloudscraper → FlareSolverr → undetected → Camoufox → Playwright → Googlebot, with premium-proxy retry. Falls back until it gets through.
bypassCloudflare: trueClean, sanitized markdown output designed to drop straight into your model's context window. Scrape any URL to text in one call.
POST /v1/scrapeExtract structured data with a JSON Schema, with optional LLM fallback and chunking for long documents. Schema.org / JSON-LD included.
POST /v1/extractAnalyze images with free OpenRouter vision models and automatic 5-model fallback. Ask a question, get an answer about the pixels.
POST /v1/visionStep-based automation for login flows and form fills, plus save-and-restore persistent browser sessions across runs.
POST /v1/automateWatch pages for change, diff two snapshots, and schedule cron monitors with webhook + Slack delivery when content shifts.
POST /v1/watch · /v1/diffParse PDF, DOCX, images (OCR), CSV and JSON into text. Scanned documents included.
POST /v1/parseSix CAPTCHA providers with automatic fallback across them — retry until the challenge clears.
POST /v1/auth/captchaWhat Pry can do
From Cloudflare-protected pages to crypto launches to local listings — one self-hosted engine with a pay-per-call layer for agents. Every card ships verified endpoints; an agent-ready x402 + MCP layer (42 tools) sits on top.
the money-printer
Multi-sportsbook odds, arbitrage + +EV, player props, injuries and live in-play — a penny a scan, pay-per-call, no subscription. A month of scans still costs less than one OddsJam subscription.
/v1/sports/odds/v1/sports/arbitrage/v1/sports/plus-ev/v1/sports/props/v1/sports/liveper scan
$0.01
pay-per-call · no subscription
Extract any public page — even behind Cloudflare, DataDome or bot walls — as clean markdown or structured JSON.
/v1/scrapeDrive real browsers through login flows and form fills; capture full-page screenshots on demand.
/v1/automate · /v1/screenshotParse PDFs, DOCX, scanned images (Tesseract OCR), CSV and JSON into clean text and structured data.
/v1/parse · /v1/ocr/extractToken launches, CEX listing signals, and rug/honeypot security checks with risk scores.
/v1/crypto/token-launches · /v1/crypto/cex-listingsAmazon, Walmart and TikTok Shop product data — price, availability, reviews and trending.
/v1/marketplace/amazon/search · /v1/marketplace/tiktok/trendingCourt dockets, SEC filings, sanctions lists and corporate registries — extracted and monitored.
/v1/legal/dockets/search · /v1/legal/sec/searcheBay, Shopify, Etsy and AliExpress product and platform data — the long tail beyond the majors.
/v1/commerce/search · /v1/commerce/platformsProperty listings, price history and parcel-level market data from major platforms.
/v1/realestate/searchJob postings across Greenhouse, Lever, Workday and other ATS systems — plus salary data.
/v1/jobs/search · /v1/jobs/salary/{company}Flight prices, hotel rates and availability — without GDS fees.
/v1/travel/flights/search · /v1/travel/hotels/ratesMulti-sportsbook odds, arbitrage + +EV, player props, injuries and live in-play — flat $0.01/scan.
/v1/sports/odds · /v1/sports/arbitrage · /v1/sports/plus-evClinical trial search, drug pricing and insurance quotes — structured, monitorable data.
/v1/health/trials/search · /v1/health/drug/{drug}SERP positions, AI Overview presence, and ad-spend intelligence.
/v1/attention/serp · /v1/seo/analyzeGoogle Maps and Yelp listings, reviews, ratings and hours.
/v1/local/places/search · /v1/local/reviews/{place_id}Procurement notices, recalls, patents and trademarks — searchable and monitored.
/v1/gov/procurement/search · /v1/ip/patents/searchExpose any domain to AI agents over MCP (42 tools), billed per call in crypto — no account, no KYC.
/v1/x402/* · MCPCrypto market intelligence
Token launches, CEX listing/delisting signals, and rug/honeypot security scans with risk scores — for sniper bots, trading agents, and security tooling.
/v1/crypto/token-launches/v1/crypto/cex-listings/v1/crypto/token-security/{chain}/{address}Agent-ready
A full MCP server over three transports, a GPT Action manifest, and a SKILL.md — plug Pry into Claude, Cursor, or Hermes in one line.
stdio
pry mcp serve — local, zero-config
HTTP
/mcp/messages/{session_id}
SSE
/mcp/sse — streaming events
one-line config
claude mcp add pry -- pry mcp serve"pry": {"command": "pry", "args": ["mcp", "serve"]}hermes mcp connect pry --stdioA ready-made skill file that teaches any agent how to scrape, extract and monitor with Pry — drop it in and go.
Expose Pry as a custom GPT Action for ChatGPT — scrape the web from inside a conversation, priced per call via x402.
Deploy once. Gate /v1/scrape behind x402. Every agent that calls it pays you — in crypto, settled on-chain.
No subscriptions. No Stripe. No KYC.
docker compose up — Pry serves /v1/scrape on your box or ours.
Wrap /v1/scrape in x402. Every call returns 402 with a price.
Agents pay per call in USDC on-chain. No account, no signup.
Facilitator verifies + settles. You receive crypto, instantly.
> POST /v1/scrape
< 402 Payment Required
x402-version: 1
accepts: USDC@base, USDC@solana,
USDT@arbitrum, DAI@polygon
> PAYMENT-SIGNATURE: eyJhbGci...
< 200 OK
PAYMENT-RESPONSE: settled
tx=0x7f3a…, block=18234102Coinbase · PayAI · Cloudflare facilitators, with EIP-7702 self-verify fallback.
/v1/scrape/v1/screenshot/v1/extract/v1/crawl/v1/stealth_browser13 chains · USDC / USDT / DAI / native assets.
Compare
The honest scoreboard. Full deep-dives on our Crawl4AI, Firecrawl, ScrapingBee and Oxylabs pages.
| Capability | pry | Crawl4AI | Firecrawl |
|---|---|---|---|
| Self-hosted | ✓MIT core | ✓Apache-2.0 | ✓open source |
| Stealth engine | ✓Camoufox + TLS + 6 JS scripts | △basic browser stealth | ✓managed anti-bot |
| Cloudflare bypass | ✓10-tier auto-fallback | △external FlareSolverr | ✓built-in |
| MCP multi-transport | ✓stdio + HTTP + SSE | △stdio | △stdio + HTTP |
| x402 crypto payments | ✓pay-per-call, no KYC | ✗— | ✗— |
| Price | $0free + pay-per-call | $0free | $free tier → subscription |
Code
Same API whether you curl it, drive it from Python, or hand it to an agent.
curl -X POST http://localhost:8005/v1/scrape \
-H "Content-Type: application/json" \
-d '{
"url": "https://example.com",
"bypassCloudflare": true,
"formats": ["markdown"]
}'Pricing
Start free on your own box. Charge per call with x402. Or let us run it for you.
MIT core. Unlimited requests on your own hardware.
Pay crypto per call. No account, no infra, no KYC.
We run it. Managed stealth, proxies, dashboard, SLA — Free, Pro, Team, Enterprise. Pay in crypto via x402 recurring — no KYC.
x402 pay-per-call · basic stealth
priority stealth · residential proxies · standard support
batch · custom templates · priority support
SLA · dedicated · unlimited
Pro $49/mo · Team $199/mo · Enterprise custom — no live checkout yet, email us.
Pay with — multi-chain, multi-asset
BaseEthereumArbitrumSolanaTRONUSDCUSDTETHARBEVM networks live today · Solana & TRON rolling out · no card required
Core is MIT. The stealth / anti-detection module is source-available under BSL 1.1 — free for personal, OSS and 90-day evaluation use; commercial license for production. See the license.
Subscribe
We run the stealth browsers, the residential proxy pool, the dashboard and the SLA — you get 12 data domains, one API, and crypto-native billing. Managed signup opens soon; email us and we'll onboard you first.
Team $199/mo · Enterprise custom · pay in crypto or card · no lock-in
FAQ
Scraping publicly available data is generally legal, but you must respect each site's terms of service, robots.txt, copyright, and privacy laws like GDPR/CCPA. Pry ships compliance tooling — consent recording, deletion requests, retention policies and audit logs — so you can build responsibly. Use it for legitimate purposes, keep rate limits civil, and don't hoover up personal data.
x402 is the HTTP 402 Payment Required standard, native to crypto. When a client hits a gated endpoint, Pry replies 402 with a price and the assets it accepts (USDC, USDT, DAI or native tokens across 13 chains). The client pays on-chain, a facilitator (Coinbase, PayAI or Cloudflare) verifies and settles the payment, and Pry returns your data. No account, no signup, no KYC — the payment itself is the credential.
Self-host if you want full control, zero cost, and data that never leaves your infrastructure. Choose the hosted subscription if you'd rather not operate FlareSolverr, browsers, and proxies yourself — we run managed stealth, a residential proxy pool, dashboards and an SLA. The x402 pay-per-call lane sits in the middle: your code, our per-call billing.
Pry uses a 10-tier fallback ladder — direct HTTP with smart headers, TLS fingerprinting, cloudscraper, FlareSolverr, undetected Chrome, Camoufox stealth Firefox, Playwright, Googlebot mimicry, and premium-proxy retry. It works on a large share of protected sites, but no scraper bypasses everything forever; specific sites and detection changes vary. The ladder is designed to try cheaper tiers first and only escalate when blocked.
The core engine is MIT open source and free to self-host, forever, with unlimited requests. The stealth/anti-detection module is source-available under BSL 1.1 — free for personal use, non-commercial open source, and 90-day evaluation, with a commercial license for production. Hosted tiers are optional. No dark patterns: the free tier is genuinely free.
Something else? Ask in Telegram or @cryptorugmunch.
One command to a self-hosted scraper your AI agents can use — and pay for — today.