How do you read an "open ports on my IP" scan result?
AdminKit's open ports on my IP result reports, for each of nine TCP ports, whether something on your public IP accepted a connection within a timeout — nothing more. Reading it means matching each open port to the device behind it, since a Telnet port left open needs a far more urgent fix than a deliberately exposed admin web UI.
By AdminKit team, Netbase JSC · Updated
Key takeaways
- “Open” means one thing: a TCP connection to that port on your public IP completed within the check’s timeout. It says nothing about passwords, versions or vulnerabilities.
- The check tests nine TCP ports from its default list of eleven. BACnet/IP (47808) and CoAP (5683) are UDP and are listed as not scanned.
- Seven of the nine are flagged high severity when open (Telnet, RTSP, MQTT, MQTTS, Modbus, TR-069, 8080). Plain web ports 80 and 443 are informational.
- Closed and filtered both show as “Closed/filtered”. The difference matters for your firewall, not for exposure.
- Fix by priority: protocols with no login step first, admin web UIs after. Re-run the check to confirm each fix took effect.
What does the result actually measure?
AdminKit’s open ports on my IP check tests nine specific TCP ports against your own public IP — the address you’re connecting from, determined server-side, never one you type in — and reports each as open or not, only after you click to start it. “Open” here has one precise meaning, consistent with how NIST’s firewall guidelines describe basic port filtering: a TCP connection attempt to that port completed successfully within the timeout, which means some process on that address is actively listening on it. The check stops there — it never sends a login attempt, a protocol command, or anything resembling an exploit, so an open result tells you a door exists, not what’s behind it or whether it’s locked.
In TCP terms, the check opens a connection, which is the three-way handshake described in RFC 9293, and closes it straight away if it succeeds. Each port gets 1.5 seconds. A service that is slow to accept can therefore show as closed/filtered even though it is listening, which is one reason to re-run a surprising result before acting on it.
What does “not open” mean: closed or filtered?
Two different outcomes both show as not-open, and the distinction is useful context even though the check doesn’t need you to act on it: a closed port means a host answered and explicitly refused the connection — there’s a live, reachable device at that address, it simply isn’t listening there. RFC 9293 makes this the normal reply: a SYN that doesn’t match any connection “is rejected” with a reset. A filtered or timed-out result means nothing came back at all, which is the signature of a firewall silently dropping the attempt rather than replying to it. If you expect a port to be closed by firewall rather than by the service simply not running, a filtered/timeout result is the one that confirms the firewall is actually doing its job; a closed result from a host that answers everything else could mean the firewall isn’t filtering that specific port at all, just that nothing happens to be running there today.
| State | What came back | Shown in AdminKit’s result | What it tells you |
|---|---|---|---|
| Open | The handshake completed | Open | Something is listening and reachable from the internet |
| Closed | A TCP reset (RST) | Closed/filtered | A host answered; nothing listens on that port |
| Filtered | Nothing, until the timeout | Closed/filtered | A firewall or router dropped the attempt |
Dedicated scanners such as Nmap separate these states and add others; the Nmap reference guide lists six, and notes that the same port can look open from one network and filtered from another at the same moment. That is why AdminKit tests from outside your network: the result describes what the internet sees, not what your LAN sees.
Which nine ports does the check test, and what does an open one hand over?
The check’s default list has eleven ports. The two UDP ones, BACnet/IP (47808) and CoAP (5683), can’t be observed with a TCP connection, so the result lists them under “Not scanned” rather than reporting them as closed. The other nine are tested, and each open one becomes a finding:
| Port | Service | Finding severity if open | Why it is on the list |
|---|---|---|---|
| 23 | Telnet | High | Unencrypted remote admin |
| 554 | RTSP | High | Camera or media stream, often unauthenticated |
| 1883 / 8883 | MQTT / MQTT over TLS | High | Broker: check for anonymous access, or that client auth is required |
| 502 | Modbus/TCP | High | No built-in authentication |
| 7547 | TR-069 (CWMP) | High | ISP device management, a frequent exploit target |
| 8080 | Alternate HTTP admin | High | Confirm it needs to be public |
| 80 / 443 | HTTP / HTTPS | Info | Often intentional, still worth confirming |
In more detail, grouped by what an open result actually hands over:
- Telnet (23) — an unauthenticated or weakly authenticated plaintext admin shell on many consumer routers and cameras. CISA’s Known Exploited Vulnerabilities Catalog lists numerous device vulnerabilities reachable precisely through exposed management ports like this one. Treat an open result here as urgent; see Telnet and RTSP: why camera and router admin ports end up exposed and is port 23 (Telnet) exposed? for the fix.
- RTSP (554) — a live video stream, often viewable with no credential check at all depending on the device’s configuration. See is port 554 (RTSP) exposed?.
- MQTT (1883) and MQTTS (8883) — a publish/subscribe broker; 1883 is unencrypted and, without an access-control list configured, frequently lets anyone publish or subscribe to any topic. See is port 1883 (MQTT) open?.
- Modbus (502) — an industrial control protocol with no authentication built in at all by design; any open result on this port in a residential or office context, rather than an isolated OT network, is worth investigating immediately. See the Modbus TCP 502 remediation playbook.
- TR-069/CWMP (7547) — a remote-management protocol ISPs use to configure routers; an open result not expected by your own ISP relationship is unusual and worth checking. See TR-069 port 7547 exposure, explained.
- Admin web UIs (80, 443, 8080) — at minimum these require a login page before anything happens, which is why they’re lower urgency than the protocols above, but an open result you didn’t intentionally configure (a router’s admin panel reachable from the internet, for instance) is still worth closing. See is port 8080 (admin panel) exposed?.
What does a full result look like?
The example below is illustrative, not a real scan: it uses 203.0.113.10 from a documentation address range. The layout, labels, summary line and fix text are the ones the check produces.
Issue found 2/9 TCP port(s) open on 203.0.113.10
23 (Telnet) Open 554 (RTSP) Closed/filtered
1883 (MQTT) Closed/filtered 8883 (MQTT (TLS)) Closed/filtered
502 (Modbus) Closed/filtered 7547 (TR-069) Closed/filtered
80 (HTTP) Closed/filtered 443 (HTTPS) Open
8080 (HTTP (admin)) Closed/filtered
Not scanned (UDP, not checked by this tool): 47808 (BACnet/IP), 5683 (CoAP)
high Port 23 is open — Telnet — unencrypted remote admin — Close this port, or restrict it to a VPN/allowlist if it must stay open.
info Port 443 is open — Close this port, or restrict it to a VPN/allowlist if it must stay open.
The badge at the top sums up the findings:
| Badge | When you see it | How to read it |
|---|---|---|
| OK | No tested port is open | Good news for these nine ports only |
| Warning | Only 80 and/or 443 are open | Confirm you meant to publish a web server |
| Issue found | At least one high-severity port is open | Work through the open ports in the order below |
| Skipped | The address was not yours to scan | The check only scans your own connection’s IP |
The address shown after “on” is the one that was scanned. If your connection reached AdminKit over IPv4, that is your IPv4 address, and your devices’ IPv6 addresses were not tested; see why IPv6 home devices need a firewall for that side.
Why can your own address be misleading?
A public IP isn’t always a one-to-one match with a single device you control. Carrier-Grade NAT, a mobile network, or a home router sitting in front of several devices can all mean the address this check scans is shared or represents more than one thing at once. If an open port doesn’t match anything you recognize on your network, confirm whether your address is actually dedicated to you before chasing the wrong device — see what is CGNAT? for how to tell.
On a dedicated address, an open port usually points to a port-forwarding rule or a UPnP mapping on your router. The router’s port-forwarding page, and its UPnP mapping table if it has one, show which LAN device each open port leads to.
What should you do with the result?
Work down the list in the priority order above rather than treating every open port as equally urgent: an unauthenticated protocol with no login step (Telnet, RTSP, MQTT, Modbus) is a more immediate risk than a web admin UI that at least requires credentials first. For each port that’s open and shouldn’t be, the fix is almost always one of three things — disable the service if you don’t need it reachable from the internet at all, change any default credentials if you do, or move remote access behind a VPN or firewall rule instead of a direct port forward. Re-run the check after making a change; a fix that doesn’t actually take effect (a firewall rule applied to the wrong interface, a port-forward rule still active elsewhere) is common enough that confirming the result changed is worth the thirty seconds it takes.
| If this is open | And you need it from outside | Fix |
|---|---|---|
| 23 Telnet | Almost never | Turn Telnet off on the device; use SSH or the vendor app instead |
| 554 RTSP | To view a camera remotely | Remove the port forward; use the vendor’s cloud app or a VPN |
| 1883 / 8883 MQTT | For remote sensors | Require credentials and TLS (8883), or reach the broker over a VPN |
| 502 Modbus | Not from the internet | Remove the forward; reach OT devices through a VPN or an outbound-only remote-access setup |
| 7547 TR-069 | Only if your ISP manages the router | Ask your ISP whether it should be open |
| 80 / 443 / 8080 | Only for a site you publish on purpose | Close the router admin UI to the WAN; harden any site you do publish |
An open port is only one slice of what’s reachable on your network, though — if one of the admin web UIs on 80, 443 or 8080 turns out to be intentionally exposed, it’s worth also evaluating that site’s HTTP security header policy strength rather than assuming a reachable login page is the end of the review.
How do you check your own IP next?
Run AdminKit’s open ports on my IP check — it only ever scans the address you’re connecting from, and only after you click to start it. If a port-forward or firewall change doesn’t seem to be taking effect, confirm your actual public-facing address first with what is my IP, since CGNAT or a changed dynamic IP can make a stale result look like a fix that didn’t work — see what your IP lookup result actually shows for what the ASN, organization and location fields in that result actually mean. If you’re checking more than one site or connection, see using port check to inventory IoT and OT device exposure for turning repeated results like this into a running exposure record rather than a one-off check. If you’d rather have an AI agent run and interpret this same check for you, see find open ports you own from your AI assistant. This check is part of AdminKit’s security basics cluster, alongside the certificate and header checks worth running on anything you expose intentionally.
This guide is general information, not a security audit of your specific network — see our disclaimer for what AdminKit's guidance does and does not commit to.
Frequently asked questions
Does 'open' mean the port is actually vulnerable?
No — 'open' only means a TCP connection attempt succeeded within the check's timeout. It says nothing about whether the service behind it uses a default password, runs outdated software, or is otherwise exploitable; the check stops at the bare connection and never sends protocol commands or credentials. Open is the starting fact to investigate, not the conclusion.
What's the difference between a closed and a filtered result?
A closed port means something answered the connection attempt and actively refused it (a TCP RST) — there's a live host at that address, just nothing listening on that specific port. A filtered or timed-out result means no response came back at all within the timeout, which usually means a firewall is silently dropping the connection attempt rather than replying. Both outcomes mean the port isn't reachable the way an open one is; the check reports them as not-open either way.
I ran this from my own network and port 8080 shows open — is that really exposed to the internet?
The check connects to your public IP from outside your network, the same way an internet scanner would, so an open result here does reflect real internet-facing exposure, not just LAN reachability. If you separately test from inside your own network against your router's LAN address, you're testing something different and will see your internal devices regardless of what's exposed externally.
My IP shows open ports I don't recognize at all — what's going on?
Your public IP can be shared: Carrier-Grade NAT (CGNAT), a mobile carrier network, or a router in front of several of your own devices can all mean the address answering isn't the single device you expect. Confirm what's actually on your network before assuming a result belongs to a device you own — see AdminKit's guide on CGNAT and dynamic-IP monitoring for how to tell whether your address is shared in the first place.
Which open port should I fix first if I see more than one?
Prioritize by what the protocol exposes with no further step needed: Telnet (23) hands out a plaintext admin shell and has been the entry point for internet-scale botnets, so treat it as most urgent. RTSP (554), MQTT (1883) and Modbus (502) typically need no credentials to read data or publish messages by default, which is next. An admin web UI on 80/443/8080 at least requires a login screen before anything happens, making it comparatively lower urgency — though still worth closing if you didn't mean to expose it.
Does a clean result (no ports open) mean my network has no other exposure?
No — this check only tests nine specific TCP ports chosen because they're commonly misconfigured on home and small-office devices; it is not a full port scan of your IP, and it doesn't test UDP services like BACnet (47808) or CoAP (5683) at all. A clean result on these nine ports is good news for exactly those nine, not a clearance for everything else that might be reachable.
Want AdminKit to watch this for you?
Free checks tell you the state today. Protective DNS and monitoring tell you when it changes.
Talk to an expert
Huy Nguyen (David), Founder & CEO, Netbase JSC
Ask whether this fix applies to your setup, or get a second pair of eyes on your DNS and mail records.