Skip to content
Last updated July 2, 2026

Dependencies

Dependencies is where you turn implicit knowledge — “the web tier depends on the primary database” — into explicit edges HostAtlas can reason about. Every declared edge shows up on the Network Map and feeds incident impact analysis.

At /dependencies the header shows how many dependencies are mapped and how many were auto-discovered. The table lists:

  • Source — the thing that depends on something else (a server, a service, or a domain).
  • Relationship — the type of link (e.g. depends on, routes to, stores data in).
  • Target — the thing being depended on.
  • Origin — auto-discovered by the agent, or manually declared.
  • Actions — delete.

Filters and quick counts break the list down by type (server-to-server, service-to-service, domain-to-server, …).

  • Add Dependency — opens a modal:
    1. Pick the Source — type (server / service / domain) and entity.
    2. Pick the Relationship — how one relies on the other.
    3. Pick the Target — type and entity.
  • Delete — remove any dependency, auto or manual. Auto edges will re-appear on the next agent scan if the underlying signal is still there.
  • Auto-discovery. Some links come for free: a domain served on a web server, a service running on a specific host, a container on a host. The agent lifts those as auto edges — you don’t need to declare them.
  • Manual declarations cover what code doesn’t reveal on its own — the app on Server A talks to Postgres on Server B, the checkout domain relies on the payments service, and so on.
  • Every edge feeds two places:
    • The Network Map — declared edges are drawn immediately.
    • Incident impact analysis — when a target has an incident, all sources that depend on it are flagged as potentially affected. This becomes the impact graph on the incident detail.
Was this page helpful?