PRIVACY
What this site records, and who else sees a lookup
What is logged
- Every lookup, from this page, the JSON API or the text API, writes one log line: the time, your IP address and what you looked up. Opening the home page is a lookup of your own address. Something with an @ in it, such as an email address, is refused, and that line records only that it had one; the web server's own access line and the request's trace, below, still carry the path as sent.
- Refused requests. When a request is blocked, rate limited or banned, the log keeps your IP address and why, with the country GeoIP puts that address in or the path you asked for where the rule turns on them.
- Failures. When a lookup fails, the log names the domain or address that failed, without yours.
- MCP tool calls are not logged, except a failed one, which is logged as above: its target, without your address.
How long it is kept
- The log file on the server,
service.log, starts afresh every day and keeps the last 7 days of old files. Older ones are deleted. - The same lines also go to the server's console log and, with a trace of each request (method, path and query, so what was looked up, plus status, timing and user-agent), to the operator's SigNoz monitoring. How long those keep them is set in the monitoring and hosting systems themselves, not by this service.
- A ban keeps your IP address, the reason and, where they apply, the path and the country until it expires: 1 hour for going over the rate limit, 24 hours for a request shaped like a vulnerability probe. A ban set by hand lasts until it is lifted.
- The rate limiter holds the times of your recent requests in memory for a few minutes. None of it is written down.
- Results are cached by what was looked up, never by who asked: registration (RDAP and WHOIS) answers for 6 hours, in memory, and subdomain lists on disk, refreshed when a request finds them older than 7 days.
What is not collected
- No cookies, no local storage, no accounts, and no analytics or advertising scripts.
- The fingerprint panel and the WebRTC leak test run in your browser. What they find is never sent to ip.1kko.com. The panel's canvas, audio, WebGL and font probes run only when you press its Measure button, and the WebRTC test only when you press its button.
Who else the server contacts
These servers see this server's address, not yours. They do learn what was looked up, and on the home page that is your own address.
- Public DNS resolvers 8.8.8.8 and 1.1.1.1: the records of the domain you look up, and the reverse-DNS name of an address.
- RDAP registries, found through IANA's bootstrap files and, for the few registries IANA omits (.ch, .de, .io, .us among them), the list bundled with the whoisit library: the domain or address, sent to the registry or regional internet registry that holds it.
- WHOIS servers on port 43: the domain, when its registry has no RDAP service or does not answer.
- The looked-up host itself: a TLS connection to its port 443, to read its certificate.
- crt.sh: the domain, and only when you ask for its subdomains.
- api.abuseipdb.com (AbuseIPDB, in the United States): an IP address you look up directly, to read how often it has been reported for abuse; your own address only when you press "Check with AbuseIPDB" on the home page; never the address a domain resolves to. Each answer is kept in memory for 24 hours, so an address is asked about at most once in that time.
A .onion or home.arpa name is refused before any of these is asked: nothing public answers for it, and a Tor hidden service's name is sent to none of them.
The server also downloads files that carry nothing about you: the GeoLite2 databases (from download.maxmind.com, cdn.jsdelivr.net), the Public Suffix List (publicsuffix.org) and IANA's RDAP bootstrap files. Finding an address on the map uses those databases on this server; nobody else is asked.
It also downloads the public lists a lookup's Reputation card checks an address against, once a day each and never for a particular lookup: Spamhaus DROP and ASN-DROP, the Tor Project's exit list, and X4BNet's VPN and datacenter lists (from www.spamhaus.org, check.torproject.org, raw.githubusercontent.com). Checking an address reads this server's copies, so none of those servers learns what was looked up. A copy older than 3 days is reported as not checked rather than used. Spamhaus DROP and ASN-DROP data are © The Spamhaus Project SLU, used under its DROP terms.
It also downloads the public datasets behind a lookup's AS card, on a schedule and never for a particular lookup: RIPE NCC's AS names, the NRO's delegation statistics and RIPE RIS's routing tables, the RPKI route validations published by rpki-client, and CAIDA's AS relationships and organizations (from ftp.ripe.net, www.ris.ripe.net, console.rpki-client.org, publicdata.caida.org). Reading an address's route or an AS reads this server's copy, so none of those servers learns what was looked up. A dataset older than 3 days (75 days for CAIDA's relationships and 200 days for its organizations, which it publishes less often) is reported as unknown rather than used. Source: RIPE NCC. RPKI data validated by rpki-client.org. The CAIDA UCSD AS Relationships Dataset and The CAIDA UCSD AS Organizations Dataset are used under CAIDA's acceptable use terms; the AS card names the date of the file it read.
It also downloads OpenINTEL's daily measurement of the public top lists (Tranco, Cisco Umbrella, Cloudflare Radar, Majestic Million and the Chrome UX Report), at most once a day per list and never for a particular lookup, from object.openintel.nl. Only the few columns of each file that hold a name and its address are fetched. An IP address lookup's "Domains on this address" panel and a subdomain list read this server's copy, so that server learns nothing about what was looked up. Names from a list older than 4 days are not shown. The research leading to these results was made possible by OpenINTEL (https://www.openintel.nl/), a joint project of the University of Twente, SIDN, NLnet Labs and SURF. The data is used under CC BY-NC-SA 4.0; an unmodified copy of OpenINTEL's terms is kept here.
Who else your browser contacts
- tile.openstreetmap.org, for the map tiles on a page that shows a map: about 15 requests on a desktop, 6 on a phone. Zooming or moving the map on an AS page fetches more tiles from the same host. OpenStreetMap sees your IP address and this site's name, but not the page's address, so not what you looked up. The tiles it serves do show which part of the map you are viewing.
- stun.cloudflare.com:3478, a STUN server, and only when you press "Run WebRTC leak test" on the home page. It sees your IP addresses: reporting which ones it sees is the test.
- Nothing else. Fonts, scripts and styles come from ip.1kko.com. Links to GitHub, OpenStreetMap, crt.sh or MaxMind are only followed if you click them.
Questions or corrections: open an issue.