Unclassified

14 · Discovery & Activation View

Discovery is what exists / what is available. Activation is what should become active for this org/runtime/context. Engine, Activation Manager, Activation Bridge, registration, and runtime config.

Two Different Questions

Question	System	Answers
Discovery	Discovery Engine	What exists / what is available across ecosystems, connectors, and capabilities?
Activation	Activation Manager + Activation Bridge	What should become active for this organization / runtime / Work context?

Discovery without activation would expose unavailable routes. Activation without discovery would miss what the connected world can actually do.

Ecosystems → Discovery Engine → Capability catalogue (available)
                                    │
                                    ▼
                            Activation Manager
                          (policy + tenant + context)
                                    │
                                    ▼
                            Activation Bridge
                          (runtime wiring → Fabric)
                                    │
                                    ▼
                   Capability Fabric (what can be done, now — governed)

Components

Discovery Engine — crawls connected ecosystems and connectors to enumerate capabilities, entities, and routes. Produces the available catalogue.
Activation Manager — decides what becomes active for a tenant, Work, or runtime — based on configuration, entitlements, and context.
Activation Bridge — wires active capabilities into the Capability Fabric runtime so Twin and agents can resolve them under governance.
Capability registration — active capabilities are registered with metadata, contracts, and authority requirements.
Runtime configuration — per-tenant/per-environment activation state that controls what the Workbench, Twin, and agents can actually see.

Why This Separation Matters

A capability can be discovered (Salesforce supports UpdateOpportunity) without being activated for tenant A, Work type B.
Activation is governed and auditable — it is not a user toggling a raw MCP tool on.
The Pipeline and Platform Execution both depend on this boundary: they never assume a capability exists without it having been discovered and activated for the current context.

Where to Go Next

What can be done, now that it's active — resolution, ADK, availability.


May it be done, for this identity, in this Work, under this policy?


Where capabilities are discovered from.

</CardGroup>