Capzy Live web data for your AI agents.
One MCP server. Your agent asks a question in plain English, Capzy pulls the answer from the live web: Amazon, Google, eBay, Walmart, Zillow and 31 more platforms. No scraping code, no proxies, no parsing.
Ask anything. Watch it answer.
Real conversations, real extracted records. Pick a company:
{ "username": "laptops4u-14" }1 credit · $0.03laptops4u-14 has 741 active listings (98.6% positive feedback, 5.6K reviews). Recent items:
Data shown is from real captured runs of these exact tools. Every platform also carries deeper endpoints (reviews, product detail, sellers, jobs) that agents discover automatically.
Connected in two minutes
Claude Code
claude mcp add capzy-data --transport http https://mcp.capzy.ai/mcp \ --header "Authorization: Bearer YOUR_KEY"
Cursor
Add to .cursor/mcp.json:
{
"mcpServers": {
"capzy-data": {
"url": "https://mcp.capzy.ai/mcp",
"headers": { "Authorization": "Bearer YOUR_KEY" }
}
}
}
Claude.ai / Claude Desktop
Settings > Connectors > Add custom connector
URL: https://mcp.capzy.ai/mcp?key=YOUR_KEY
Any MCP client speaking Streamable HTTP works the same way.
373 skills across 36 platforms
Search is built in for every platform. Deeper endpoints (sellers, reviews, product detail, places, jobs) are discovered by your agent automatically.
Advanced: smaller toolsetstrim context by mounting less
| Server URL | What it carries |
|---|---|
/search/mcp | Only the search skill of every platform. |
/<platform>/mcp | One platform's full typed toolset, e.g. /amazon/mcp. |
/mcp?platforms=amazon,ebay | Typed tools for a hand-picked set of platforms. |
Proxies, bot defenses, retries and parsing are handled behind every tool. Your agent gets clean structured records.
A run costs its listed credits no matter how many records come back. No surprise bills, and discovery is always free.
Every run lands in your capzy.ai dashboard with the full result and csv / json / xlsx export.