# AgentReader

Pay-per-call content extraction API for AI agents (x402 protocol, USDC).

- GET https://agentreader.dev/read?url=<page-url> - main article as clean markdown ($0.005/call)
- GET https://agentreader.dev/meta?url=<page-url> - page metadata: OG tags, JSON-LD, headings ($0.002/call)
- POST https://agentreader.dev/clean - JSON body {"html": "..."} converted to markdown ($0.003/call)
- GET https://agentreader.dev/pdf?url=<pdf-url> - full PDF text extraction ($0.01/call)
- GET https://agentreader.dev/feed?url=<feed-url> - RSS/Atom feed to normalized JSON ($0.003/call)
- GET https://agentreader.dev/sitemap?url=<sitemap-url> - XML sitemap to URL list ($0.003/call)
- GET https://agentreader.dev/links?url=<page-url> - all page links, classified ($0.002/call)
- GET https://agentreader.dev/dns?name=<domain>&type=A - DNS records as JSON ($0.001/call)
- GET https://agentreader.dev/screenshot?url=<page-url> - PNG screenshot from a real browser ($0.02/call)
- POST https://agentreader.dev/summarize - JSON {url|text|html, style?, maxWords?} to AI summary ($0.02/call)
- POST https://agentreader.dev/extract - JSON {url|text|html, schema, instructions?} to structured JSON matching your schema ($0.03/call)

No API key needed. Unpaid requests return HTTP 402 with x402 payment requirements.
Payments settle in USDC via the x402 protocol (eip155:8453).

OpenAPI: https://agentreader.dev/openapi.json
x402 discovery: https://agentreader.dev/.well-known/x402
MCP server (all endpoints as agent tools with auto-payment): https://github.com/ajanani1/agentreader-mcp