The system catalog
One screen listing every source system the product knows how to read from, by category, and what the connected status actually means.
The catalog lives under Settings → Integrations and connections, and it currently lists 96 systems in fourteen categories. Each row says what that system would give the product, what kind of interface and authentication it requires, and what state it is in here. The catalog map at the top of the screen prints the current counts.
Two directions, and they must not be confused
| Direction | What it is | Where it is managed |
|---|---|---|
| Outbound — the product calls out | Credentials the product uses to read data from other systems | This catalog |
| Inbound — something calls the product | Access tokens an external system presents in order to read from here | The access tokens page |
These are opposite things that tend to be called the same word, so the head of the screen states the distinction and links to the other page.
Fourteen categories
| Category | What it holds |
|---|---|
| Virtualization | Hosts, clusters and virtual machines — what runs on what, and when a host can be taken down for maintenance |
| Monitoring and observability | Availability, performance and traffic relationships between systems — the basis for detecting collisions and for a safe window |
| Patch and endpoint management | What is installed, what is pending, and which update group each host belongs to |
| Vulnerability scanning | Which CVEs were found on which hosts, and which patch closes them |
| CMDB and ticketing | The configuration register, system owners, and the change approval process |
| Network security and firewalls | Security device inventory, firmware version and HA pairs — devices that need a window too |
| Network management and IPAM | IP addresses, site and network segment assignment, and DNS records |
| Identity and directories | Who the users are, who the system owner is, and where the credentials for performing the update are held |
| Backup and restore | When a host was last backed up — the practical precondition for any significant update |
| Storage | Storage arrays and their firmware versions |
| Cloud | Machines and managed services outside the data center |
| Databases | Instances, versions and ownership — what goes down when the host beneath them is patched |
| Ticketing, mail and messaging | Where outbound messages go: a request, a calendar invitation, a text message or an alert |
| Automation and configuration | The tools that actually perform the update scheduled here |
Five states
The order is the order the legend and the filter render them in, most configured first:
| State | What it says |
|---|---|
| Connected | The product has wiring for it and the configuration is complete. For a collection source configured on this installation it also means the last run succeeded |
| Configured | A collection source exists with credentials accepted, but the last run has not succeeded yet — or it has not run |
| Ready to connect | The product ships a connection path for this system: a credential form, and either a declared field mapping or dedicated wiring. Nothing has been configured yet |
| Custom mapping available | The product speaks the protocol and can hold the credential, but no field mapping has been written for this system. You set it up as a custom source |
| In the catalog | The system's interface shape or authentication shape is not built in the product yet, so there is no connection path to it. The row says what is missing and links to the vendor's documentation |
The last state is called "in the catalog" and not "planned" on purpose. "Planned" would be a roadmap promise, and there is none here — only the statement that the system is known and has no connection path yet.
The two middle states are not degrees of one thing: "ready to connect" means a path was written in the product for this specific system, and "custom mapping available" means it was not — but an administrator can write the mapping themselves.
"Configurable here" — a mark, not another state
Some rows have a configuration form and a connection test inside the product, and those are marked. The mark cuts across all five states and is not one of them: a row without it can still be connected — through the deployment's own configuration, which the product falls back to when no active connection exists. The same mark is the filter that narrows the list to what can be set up right now, instead of a second screen.
What a row shows, and what it never shows
An expanded row shows what this gives us — what the system would supply, stated in the product's terms rather than as the vendor's feature list — alongside the interface kind, the authentication kind, and a link to the vendor's documentation where one exists. A system whose vendor interface was not verified against official documentation in the pass that added it is marked not verified: that is a statement about what we know, not about the product on the other side.
Availability on a disconnected network
A row with a known reason it would be hard to use on a disconnected deployment carries a sentence saying it — "cloud only", "works only if the gateway itself is installed on-premises". A row without that sentence is one where nothing is known to stop it, not a promise that it will work. See Air-gapped deployment.
Finding a row
Free-text search by system name, vendor or the data it supplies; filtering by state or by "configurable here"; category selection from the catalog map; and a list or card layout, whichever reads better. The layout choice is stored on your own machine and is not a team decision.
The categories already wired into the product are described in Supported integrations.
Updated
This page is the file content/docs/en/v1/integrations/catalog.mdx