Terrantula + Atmos

Keep Atmos. Stack inheritance, component composition, DRY configuration for hundreds of similar deployments — that's Atmos's strength, and Terrantula doesn't replicate any of it.

How they compose. Two integration points. On the way in, terrantula import-atmos reads your stack manifests to populate the fleet graph — your existing Atmos structure becomes visible as entities and relationships. On the way out, Terrantula Actions invoke Atmos workflows for execution (via a customer-deployed reference runner), or open a PR that extends a stack manifest, which your existing pipeline applies.

What Terrantula adds that Atmos doesn't have: an entity model over live state. Atmos describes what stacks should exist; Terrantula tracks what does exist — capacity, lifecycle state, placement, cardinality. Atmos says "run these components in order"; Terrantula answers "is this allowed, where should it land, and what state is everything in afterward."

Zero SaaS required. Terrantula OSS + Atmos + your own CI is the complete workflow with no subscriptions.

Not for you if your Atmos estate is a few hand-tended stacks with no per-tenant or per-account stamping.

→ Atmos on-ramp · Examples gallery