> ## Documentation Index
> Fetch the complete documentation index at: https://docs.integratewise.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# 15 · Capability Fabric View

> What can be done? Capability catalogue, resolution, metadata, MCP, native capabilities, ADK, external capabilities, and availability.

## What Can Be Done?

The Capability Fabric is the **provider-neutral catalogue of what the system can do** — the answer to *what can be done?* before governance answers *may it be done?* and Hermes answers *execute it*.

```text theme={null}
             Discovery + Activation
                     │
                     ▼
           CAPABILITY FABRIC
        ─────────────────────────
        Catalogue · Metadata
        Resolution · ADK · MCP
        Native + External
                     │
            ┌────────┴────────┐
            ▼                 ▼
     Twin / Agents       Governance
      (propose)        (may it be done?)
                           │
                           ▼
                        Hermes
```

<Callout type="info">
  Capabilities are **provider-neutral**. `UpdateRenewalRisk` is a capability; `PATCH /salesforce/opportunity/{id}` is a route. The Fabric resolves the former to the latter only when active and authorized.
</Callout>

## What Lives in the Fabric

* **Capability catalogue** — the full set of active capabilities for a tenant/runtime/Work context.
* **Capability metadata** — contracts, inputs/outputs, required authority, evidence shape, side-effect class.
* **Resolution** — maps a capability name to one or more execution routes (native, MCP-routed, ADK-resolved).
* **MCP** — governed tool/context surface exposed to AI, routed through the MCP Pool.
* **ADK** — agent development kit path for agent-authored capabilities.
* **Native + external** — first-party capabilities and those contributed via ecosystems/Marketplace.
* **Availability** — not every capability is available everywhere; Discovery + Activation + tenancy + Work context determine visibility.

## Resolution Is Contextual

A capability resolves differently depending on:

* Which ecosystems are connected and authorized for this tenant
* Which routes are active for this Work type
* Which identity is requesting (user, Twin, agent — with what authority)

That is why planning happens **after** context assembly and **before** governance — the plan's capabilities are already contextual.

## Fabric ≠ Governance ≠ Hermes

```text theme={null}
Capability Fabric          Governance                Hermes
 "What can be done?"  →  "May it be done?"  →  "Execute it."
  catalogue +           policy + approval       authorized
  resolution            → authorization          execution
```

## Where to Go Next

<CardGroup cols={2}>
  <Card title="Governance & Authority" icon="shield-check" href="/platform/governance-authority-view">
    How the plan is approved or denied before anything executes.
  </Card>

  <Card title="Hermes Execution" icon="bolt" href="/platform/hermes-execution-view">
    How the authorized plan is executed with evidence.
  </Card>

  <Card title="Discovery & Activation" icon="magnifying-glass" href="/platform/discovery-activation-view">
    What makes a capability exist and become active here.
  </Card>
</CardGroup>