FFraud
AI ajanları için ücretsiz IP, e-posta ve alan adı dolandırıcılık kontrolleri: dolandırıcılık puanı, VPN, proxy ve Tor tespiti, geçici e-posta tespiti. Anahtar gerekmez.
Barındırılan MCP Sunucusu
npx add-mcp 'https://ffraud.com/api/mcp'Claude Code, Codex, Cursor ve daha fazlasına kurulur
Dokümantasyon
MCP server
Add FFraud to Claude, Cursor or any MCP client with one address. Five tools check a whole website's security, IPs, emails, domains and abuse reports. Free, no key, nothing to install.
FFraud runs an MCP server at https://ffraud.com/api/mcp. Add that one address to an AI client and the model checks a whole website, an IP, an email or a domain by itself, in a chat or inside an agent's loop. Nothing to install and no key.
Add the MCP server to your client
claude mcp add --transport http ffraud https://ffraud.com/api/mcp
In Claude's own apps, add it as a custom connector with the same address. Any other client that speaks MCP over HTTP takes it too.
A client that only runs local servers starts the same five tools from npm: npx -y ffraud mcp. The ffraud package is a command line tool too: npx ffraud site example.com, npx ffraud ip 185.220.101.34.
Step by step, with three prompts that work: Claude Code, Cursor, Codex and ChatGPT.
The seven MCP tools
| Tool | What it answers | Input |
|---|---|---|
| site_check | A whole website's security in seconds: the certificate, the security headers, cookies, http to https, SPF, DMARC and MX, the domain and its addresses on abuse lists, the bot protection the page loads. A 0 to 100 score, a grade from A to F, and every finding with the exact fix | site: a domain or any URL of the site |
| ip_check | A 0 to 100 fraud score and risk band; VPN, proxy, Tor, relay or datacenter; the network, location and abuse history; how much of its /24 is flagged | ips: one to 20 addresses |
| email_check | Disposable or throwaway, a role address, on the community blacklist, or on a domain that hands out temporary mailboxes | email: a full address |
| domain_check | Whether a domain hands out throwaway mailboxes or appears in the phishing data, with a risk level | domain: a bare domain |
| ip_reports | The community reports behind a score: what the address was reported for, when, and by how many reporters | ip: one address |
| request_api_key | Gets the person a free API key: FFraud emails them one Approve button, and nothing is created until they press it. Answers a signup_id and a claim_token | email: the person's own address; agent: your name, shown in the email |
| claim_api_key | Picks up that key once the person approved: pending while they have not, approved with api_key exactly once, or denied, expired or delivered | signup_id and claim_token from request_api_key |
Each tool answers with the same JSON as FFraud's API: the IP check, the email check, and GET /public/site/{domain} for site_check (in the OpenAPI spec). The model reads the full answer, reasons and fixes included.
Ask the model in plain words
- How secure is my website, example.com? Give me the fixes, the most important first.
- I added the headers you gave me. Check example.com again.
- Is 185.220.101.34 safe to let sign up?
- Check these five IPs from my login logs and tell me which ones to block.
- Is newuser@mailinator.com a throwaway email?
- What has 45.148.10.121 been reported for?
- Get me a free FFraud API key for me@example.com.
Keys and limits
No key is needed. The server asks FFraud's public endpoints under your own address, so the free limits are the same as on the website. A free key raises them: send it as an X-API-Key header, which the server passes on.
No key yet? Ask the model for one. request_api_key has FFraud email you one Approve button; the page it opens shows which agent asked, for which address and from where, and a new address chooses its password there. After your click, claim_api_key hands the key to the model exactly once, and it appears on your API keys page as Agent: <name>, where deleting it cuts the agent off. The same works without MCP: AI agents.
site_check reaches out to a real website, so it has an hourly allowance: 20 fresh checks per address without a key, 300 per account with one. A site checked in the last ten minutes answers from memory and costs nothing. After you change the site, ask for a fresh check (fresh: true): it runs again unless the site was checked in the last minute.
claude mcp add --transport http ffraud https://ffraud.com/api/mcp \
--header "X-API-Key: ff_live_your_key_here"
The five checks only read; site_check reads only what any visitor of the site sees. The two key tools act, and say so to your client, so a client that confirms actions asks you first: request_api_key sends you one email, and a key exists only after your click. No tool can spend money.
Test the server with curl
The server speaks JSON-RPC 2.0 over HTTP POST and answers with one JSON message, with no session to open first. This asks for the tools, then calls one:
curl -s https://ffraud.com/api/mcp \
-H "Content-Type: application/json" \
-d '{"jsonrpc":"2.0","id":1,"method":"tools/list"}'
Clients and registries that discover servers by domain find FFraud at https://ffraud.com/.well-known/ai-catalog.json, which points to the server card at https://ffraud.com/api/mcp/server-card (MCP Server Cards, SEP-2127): the server's name, version and address, read before connecting.
What is an MCP server?
The Model Context Protocol is a standard way to hand an AI model tools it can call. Claude, Cursor, VS Code and other clients speak it, so one address gives the model FFraud's checks.
Does the MCP server cost anything?
No. It is free like the rest of FFraud and needs no key. Only site_check has an hourly allowance, because each check reaches out to a real website.
What does FFraud see when the model calls a tool?
The IP address, email or domain the call sends, the same as a check on the website. The conversation itself stays with your client.
Can the model block an address or change my account through it?
No. All five tools only read.