Context, Connections & Providers
Feed live STRATUM context and external connection references into the Operator DAG without turning credentials into prompt text.
Give the Operator authoritative data, not guesses
Context nodes connect the Operator DAG to the world STRATUM is actually managing. The goal is to keep infrastructure state, dependency evidence and provider connections explicit in the graph so the AI can reason from controlled inputs instead of inventing environmental facts.
Connection nodes represent reusable provider/cluster/cloud connectivity. They can be consumed by Skills that need an external API or managed system. Credentials remain managed by STRATUM; the DAG works with the connection object rather than exposing secret material as graph text.
Context and connection nodes
Context & Connections
- STRATUM Infrastructure — Injects live STRATUM topology and infrastructure state into the Operator so planning can reason from the datacenter that actually exists.
- STRATUMops — Supplies relationship, dependency, lineage and operational evidence from STRATUMops so a Mission can reason across systems instead of at a single node in isolation.
- Cloud Context — Describes the cloud environment in which an Operator is reasoning: platform, site and environmental context without making the context itself a credential.
- Cloud Connection — References a configured cloud connection that Skills or provider-aware operations can consume.
- Cluster Connection — References a Kubernetes/OpenShift/RKE2/Rancher-style cluster connection for cluster-aware Skills.
- Provider Connection — Provides a reusable typed connection to an external provider or managed system. The graph carries a connection reference; secrets remain managed by STRATUM rather than copied into prompts or canvas labels.
AI provider and Operator context are separate
STRATUM AI Administration selects the AI provider/model and its compute limits. AI Studio defines the Operator that uses that model. Context/Connection nodes are therefore about infrastructure and service access, not about choosing GPT, Gemini, llama.cpp or another model in the graph.
This separation keeps the DAG portable: an Operator can preserve its Roles, Skills, guardrails and verification rules even when the underlying model endpoint changes.