About agentreader-mcp
agentreader-mcp is an MCP server offering web research, sanctions screening, and x402 trust tools to AI agents. Calls are paid in USDC on Base through x402, without API keys or accounts.
Key features
- 01Web page extraction and metadata
- 02PDF, feed, and sitemap parsing
- 03Screenshots and structured JSON extraction
- 04DNS lookup and link extraction
- 05OFAC sanctions screening
- 06x402 service trust lookup
Use cases
- 01Turn web pages into clean markdown
- 02Extract text from PDFs and feeds
- 03Screen names or crypto addresses against OFAC lists
- 04Check verified-delivery histories for x402 services
MCP support
It exposes 13 pay-per-call tools to MCP clients, spanning web content extraction, DNS, sanctions screening, and x402 trust data. Payment is handled automatically in USDC on Base through x402; failed requests are not charged.
Configuration
Local server config is the same for every client. Copy it into your client’s MCP config file.
npx -y agentreader-mcpAGENTREADER_PRIVATE_KEYSecret0x-prefixed private key of a wallet holding USDC on Base mainnet (use a dedicated agent wallet)AGENTREADER_APIAPI base URL override (default https://agentreader.dev)Local servers need a runtime on your machine (e.g. Node.js 18+). Config file locations differ by client; see your client’s docs.
Tools
Local servers expose tools only once running on your machine, and the maintainer has not provided a tool list.
Details
From the project READMEagentreader-mcp
MCP server for AgentReader — 13 pay-per-call tools for AI agents (web intelligence, OFAC sanctions screening, x402 trust data), paid automatically in USDC on Base via the x402 protocol. No API keys, no accounts: your agent's wallet pays $0.001–$0.05 per call and failed requests are never charged.
Tools
| Tool | Price | What it does |
|---|---|---|
read_page | $0.005 | Web page → clean LLM-ready markdown article |
page_metadata | $0.002 | Title, OpenGraph, Twitter cards, JSON-LD, headings |
clean_html | $0.003 | Raw HTML → markdown |
pdf_text | $0.01 | PDF URL → full text (up to 25 MB) |
parse_feed | $0.003 | RSS/Atom/RDF feed → normalized JSON items |
parse_sitemap | $0.003 | XML sitemap → URL list |
extract_links | $0.002 | Page link graph with anchor text |
dns_lookup | $0.001 | DNS records with TTLs |
screenshot | $0.02 | Real-browser PNG screenshot |
summarize | $0.02 | AI summary of URL or text (dense, factual) |
extract_structured | $0.03 | Page/text → JSON matching your JSON Schema |
sanctions_screen | $0.05 | OFAC SDN + Consolidated screening: fuzzy transliteration-aware name match or crypto-address lookup, with verdict + list provenance |
x402_trust_lookup | $0.05 | Verified-delivery history for any x402 service, built from real paid test purchases (free aggregate: /leaderboard) |
Setup
You need a wallet holding a little USDC on Base mainnet ($1 covers hundreds of calls). Use a dedicated agent wallet, not your main one.
Claude Desktop / Claude Code / any MCP client:
{
"mcpServers": {
"agentreader": {
"command": "npx",
"args": ["-y", "agentreader-mcp"],
"env": { "AGENTREADER_PRIVATE_KEY": "0xYourAgentWalletPrivateKey" }
}
}
}
Also listed in the official MCP Registry as io.github.ajanani1/agentreader-mcp and on npm.
Environment variables:
AGENTREADER_PRIVATE_KEY(required) — private key of the paying wallet (USDC on Base; gas not needed, x402 uses gasless USDC authorizations)AGENTREADER_API(optional) — API base URL, defaulthttps://agentreader.dev
How payment works
Each tool call hits the AgentReader API, receives an HTTP 402 with x402 payment requirements, signs a USDC authorization with your key, and retries. Settlement happens through the Coinbase CDP facilitator on Base mainnet. Errors (4xx/5xx) never settle — you only pay for delivered results.
API without MCP
The same endpoints are callable directly over HTTP with any x402 client: catalog · OpenAPI · llms.txt · x402 discovery
License
MIT
Comments
New to MCP?
From what the protocol is to installing your first server, these answers cover most first questions.
An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.