Coming soon
LabGaze
An inventory for the network you run. It finds the hosts on a private subnet and keeps a host together when the address changes.
Private until it is ready. Free for one person’s own non-commercial use. If it takes payment, it needs a license.
What it is
LabGaze is the list of machines on a network you administer. You pick a private subnet. It pings that range and keeps the hosts that answer. A sweep stays inside a private network, and only one runs at a time. Virtual bridges from containers are shown and are not offered as something to scan.
A host with a real hardware address stays one host when it is handed a new address. The notes and the port history move with it, and the earlier address stays on the record. A name the next sweep does not see is not thrown away.
Optional probes add a short list of services, and a note when something is open that usually should not be, such as Telnet, SMB, or Redis. Checks for known flaws stay on hosts you administer. They can look at a web response, a TLS handshake, a version banner, a package version from a login you provided, or a read-only look at how the machine is configured. A larger set of public detection templates can be added, and only the simple web checks are used. Guessing passwords and calling out to the internet are not part of it.
You can ask a local model about the inventory, through the same tools the page uses. Free for one person’s own non-commercial use. Any use that takes payment needs a license. The repository stays private until it is ready.
First version
- Scan a private IPv4 range, or a single address. The suggested networks come from the machine’s own interfaces.
- Optional probes of common ports, with a note for services that are usually a bad idea to leave open.
- One record per hardware address, so a new address does not split the host.
- A view of how a host is being used, separate from the discovery sweep.
- One-shot and scheduled checks: web, TLS, banners, and credentialed package and config facts.
- An optional local checkout of public detection templates, limited to simple web requests. Raw requests, payloads, and login guesses are skipped.
- Findings on the host and on a list of advisories.
- The same inventory tools for a local model, in chat and over MCP.
After that
- Device labels you attach yourself, kept as the product’s own record.
- Uptime, as its own history rather than a single look at the host.