Give Claude Code and Codex a web scraping skill
Drop the Scrappey skill into Claude Code or any SKILL.md-compatible agent and it gains a web data capability that returns clean HTML or LLM-ready Markdown from JavaScript-heavy and modern websites. Prefer the terminal? The scrappey-cli scrapes any URL from a single command, with managed sessions and residential proxies handled for you.
Quick start
Set your key once with SCRAPPEY_API_KEY, or run scrappey-cli auth --api-key YOUR_API_KEY.
npm install -g scrappey-cli- 1
Add the skill to your agent
Clone the skill into your Claude Code skills directory. It ships a SKILL.md that is auto-discovered, so no slash command or extra wiring is needed. Use a personal install for all projects, or drop it into a single project.
bash# Personal install (all projects) git clone https://github.com/pim97/scrappey-skill ~/.claude/skills/scrappey # Or project-only git clone https://github.com/pim97/scrappey-skill .claude/skills/scrappey - 2
Set your Scrappey API key
Both the skill and the CLI read the SCRAPPEY_API_KEY environment variable. Register for a key and export it in your shell.
bash# macOS / Linux export SCRAPPEY_API_KEY="YOUR_API_KEY" # Windows PowerShell $env:SCRAPPEY_API_KEY="YOUR_API_KEY" - 3
Let the agent scrape, or do it yourself
Once the skill is installed, just ask Claude Code to fetch a page and it will route the request through Scrappey. From the terminal, the CLI scrapes any URL and can return Markdown directly.
bash# Clean HTML scrappey-cli scrape https://example.com # LLM-ready Markdown saved to a file scrappey-cli scrape https://example.com --markdown --output page.md
Use it from the CLI or the raw API
# Markdown output, full JSON envelope, and a specific proxy country
scrappey-cli scrape https://example.com/products \
--markdown \
--country us \
--output products.md
# Check your account balance
scrappey-cli balance
# Create a managed session to reuse across requests
scrappey-cli session createcurl -X POST "https://publisher.scrappey.com/api/v1?key=YOUR_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"cmd": "request.get",
"url": "https://example.com",
"markdown": true
}'Why pair Scrappey with your coding agent
Auto-discovered skill
The SKILL.md is picked up by Claude Code and SKILL.md-compatible agents automatically. No slash command, no manual tool registration.
Clean HTML or Markdown
Get raw HTML or LLM-ready Markdown from JavaScript-heavy and modern websites, so your agent reasons over content instead of fighting page loads.
Web access handled for you
Residential proxies and automatic web access handling are built in, giving high success rates without extra proxy setup or billing.
Works in the terminal too
scrappey-cli scrapes any URL from a single command, with flags for Markdown, output files, HTTP method, and proxy country.
Managed sessions
Create and reuse sessions across requests to keep state when collecting multi-step, publicly available data.
Pay only for success
A free trial to start, then pay-as-you-go with no subscription. Failed requests are free, so you only pay for data you actually get.
Popular use cases
Agentic research
Let Claude Code or Codex pull current, publicly available pages mid-task and feed clean Markdown straight into its reasoning.
Terminal data pulls
Grab a single URL or batch of pages from the CLI for quick checks, scripts, or piping into other tools.
Building data pipelines
Prototype scrapers with the agent, then move to the raw API call once you know the shape of the data you need.
Claude & Codex FAQ
How do I add the skill to Claude Code?
Do the skill and CLI use the same API key?
How do I get LLM-ready Markdown instead of HTML?
What does it cost?
Does it work with OpenAI Codex and other agents?
Can it handle JavaScript-heavy sites?
More ways to plug Scrappey into your stack
Start building with Scrappey
Try it free. No subscription, no credit card, instant setup — your trial is waiting.