VerifIP
VerifIP is an IP reputation and fraud-scoring API that answers one server-side HTTPS call with a 0–100 fraud score, an allow / challenge / block verdict, and VPN, proxy, Tor and datacenter flags, along with the signals that decided it. One request tells you who's calling. It also scores emails, phone numbers and URLs.
What does the IPreputation API return?
One call returns every signal below. In other words, use the fields you need and ignore the rest; either way, you are billed by request, not by signal.
IP Reputation
Blocklist consensus, botnet C2 (command-and-control) and malware-host listings — correlated feeds counted once, not stacked.
ASN Intelligence
The ASN (autonomous system number) behind each IP, with hosting and datacenter networks, VPN, proxy and Tor each flagged in its own field.
VPN and proxy detection
Commercial VPN ranges, open proxies, data-center ranges — each flagged in its own field.
Tor exit node detection
Tor exit nodes, FireHOL proxy and anonymizer lists, iCloud Private Relay.
Bot Classification
Verified Googlebot/Bingbot separated from scrapers and scanners.
Geolocation
Country, region, city and timezone from MaxMind GeoLite2.
Blocklists
Spamhaus DROP, FireHOL, blocklist.de, CINS, GreenSnow, IPsum, Emerging Threats — grouped so one fact is never counted twice.
Verdicts
allow / challenge / block on every answer. Classification alone never challenges — it takes an accusation.
What does a VerifIPresponse look like?
A JSON answer you can act on: every field explainable, every decision auditable. Log signal_breakdown next to each decision, so every block can be explained later.
// GET /v1/check?ip=185.220.101.1 (illustrative, from the API guide) // The score is not the sum of the breakdown: correlated signals are combined, not stacked. { "ip": "185.220.101.1", "ip_version": 4, "fraud_score": 43, "verdict": "challenge", "threat_confidence": "low", "threat_categories": ["tor", "proxy", "attacker"], "is_proxy": true, "is_vpn": false, "is_tor": true, "is_datacenter": true, "is_known_attacker": true, "is_botnet_c2": false, "cloud_provider": null, "is_good_bot": false, "blocklist_count": 3, "country_code": "DE", "isp": "Stiftung Erneuerbare Freiheit", "asn": 60729, "connection_type": "Data Center", "signal_breakdown": { "tor_exit": 25, "proxy_detected": 15, "datacenter_ip": 10, "blocklist_consensus": 12, "known_attacker": 15, "firehol_l2": 15 } }
Can I check up to100 IPs in one request?
Yes, on paid plans. When you're processing logs, reviewing sign-ups or screening a bulk import, the batch endpoint removes the overhead of per-IP round trips without any change to pricing.
Up to 100 IPs per request
Score a full page of users, a log batch, or a sign-up queue in a single HTTP call. No looping required.
Results preserve input order
The i-th result always corresponds to the i-th input IP. Zip against your source array without keying by IP.
Same pricing as single calls
Each successfully scored IP counts as one request toward your quota; invalid or unscored entries do not. No batch surcharge, no minimum.
One round trip, not a hundred
One HTTP round trip for up to 100 IPs, instead of one per IP.
Where does VerifIP’sthreat data come from?
Tracked, freshness-checked feeds
Tor, FireHOL, blocklist.de, IPsum, GreenSnow, CINS, Emerging Threats, Spamhaus DROP. Every source is listed on the Data sources page.
Botnet & malware
Botnet command servers and malware hosts, flagged in their own fields.
Cloud & bot detection
AWS, Google Cloud, Cloudflare, Fastly ranges + Googlebot, Bingbot identification.
Proxy intelligence
X4BNet VPN ranges, FireHOL and iplocate proxy lists, Tor exit nodes, iCloud Private Relay.
Consensus scoring
Blocklist count across five independent lists — correlated feeds are grouped, so one listing is never counted twice.
Verified crawlers
Googlebot and Bingbot, matched against the IP ranges Google and Microsoft publish.
How fast doesVerifIP answer?
VerifIP runs on Cloudflare Workers and answers in one call. If we can't read the threat data, we say so: a 503 marked X-VerifIP-Degraded with no verdict, so your code decides what happens next.
Latency distribution · typical
VerifIP questions,answered
What teams ask before they put an IP reputation check in the request path.
VerifIP’s fraud_score is a number from 0 to 100 that rates how risky an IP address is. Correlated signals are combined, not stacked, so the score is not the sum of signal_breakdown. Enforce on verdict (allow, challenge or block) and use the score for ordering.
Yes. Each answer flags is_vpn, is_proxy, is_tor and is_datacenter separately. The sources are commercial VPN and datacenter ranges (X4BNet), open-proxy lists (FireHOL, iplocate, and IP2Proxy LITE via FireHOL) and the Tor Project’s exit list. The Tor exit list refreshes every 6 hours; the others daily. iCloud Private Relay is reported as is_icloud_relay.
No. Tor, VPN, proxy and hosting are classification, not accusation: on their own they top out at 40, which is still allow. A challenge needs accusatory evidence such as attack-feed or blocklist listings, and a block needs several accusations together with an anonymity classification (or a bogon). If you want to refuse all VPN traffic, read is_vpn in your own code.
You get an HTTP 503 marked X-VerifIP-Degraded with no verdict, never a guessed one. A degraded call is billed once, and so is each retry. The docs recommend retrying once or twice with back-off and then failing closed. A 503 capacity_unavailable means nothing ran and is not billed.
Yes. /v1/check, /v1/check/batch, /v1/whois, /v1/assess (its ip parameter) and /v1/report accept IPv6 addresses. IPv6 answers list the sources with no IPv6 data in signals_unavailable, and blocklist_count is always 0 for IPv6.
Yes, on paid plans. The batch endpoint takes 1 to 100 items, returns results in input order, and bills only the entries it answered. On the Free plan a batch is refused with 403 plan_required, and that refusal is not billed.
The fastest feeds, the Tor exit list among them, refresh every 6 hours. Most blocklists refresh daily, and some range lists refresh weekly. When a feed behind an answer is older than its freshness budget, the answer names it in stale_sources.
Email addresses (including disposable-domain detection), phone numbers (validation and line type), URLs (including PhishTank listings) and WHOIS for an IP address’s network registration, looked up via RDAP (the Registration Data Access Protocol). /v1/assess checks several of these in one call.