Marketplace Extensibility & AI Packs

STRATUM AI is extensible through Skills and AI Packs. Marketplace distributes reusable operational capability; Arsenal stores the downloaded definitions and versions them; AI Studio turns those definitions into configured Skill instances inside an Operator DAG.

That lifecycle is deliberate: Marketplace → Arsenal → AI Studio → Operator. Downloading a Skill makes it available. Adding and configuring it in AI Studio decides how that capability participates in the Operator.

Drag a Skill onto the canvas and connect its Capability output to a compatible specialist or planning input. The edge is not decorative: it means that Skill is contributing operational capability to that part of the Operator.

Right-click a Skill to configure the current instance or customize its Skill definition. Instance configuration lets one Operator use the same reusable Skill differently from another. Definition customization changes the reusable capability itself.

Skills can declare configuration fields that AI Studio renders for the operator. A networking Skill might expose capture duration or convergence checks; a deployment Skill might expose artifact choices or verification requirements. That turns a downloaded Skill into a governed, operator-specific capability rather than a static JSON file sitting in Arsenal.

AI Packs group Skills and related operational content for distribution. They provide a clean way to ship new capability to a STRATUM environment without baking that capability into the base application image.

A Skill does not have to contain its own model or “AI code” to be valuable. The power is in how it exposes tools, configuration, planning guidance, policy expectations and verification to the Operator. The LLM supplies reasoning; the Skill makes that reasoning operational and repeatable.