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

# Capabilities: Governed Actions Twin and Agents Can Execute

> Capabilities are the discrete governed actions Twin, agents, and humans can execute against source systems, each with scope, authorization, and receipts.

Capabilities are the vocabulary of action in SpineWorkspace. Every write to a source system, every message sent, every approval, and every workflow step is expressed as a capability, with a declared scope, authorization requirement, and receipt.

## What a Capability Includes

* **Scope** — which entities and systems the capability touches.
* **Inputs and outputs** — declared shape of the action.
* **Authorization** — who or what can invoke it and under which policy.
* **Receipt** — the audit record produced after execution.

## Kinds of Capabilities

<CardGroup cols={2}>
  <Card title="Read capabilities" icon="magnifying-glass">
    Assemble context, fetch records, and inspect state.
  </Card>

  <Card title="Write capabilities" icon="pen">
    Update records, create entities, and post communications.
  </Card>

  <Card title="Communication capabilities" icon="envelope">
    Send email, chat, and other messages under governed identity.
  </Card>

  <Card title="Workflow capabilities" icon="diagram-project">
    Kick off workflows, approvals, and multi-step automations.
  </Card>
</CardGroup>

## Related

* [Governed Actions](/how-it-works/governed-actions) — how capabilities are approved and audited.
* [Agents](/agents/spine-agent-network) — who invokes capabilities.
* [Governance](/security-and-governance/governance) — the policy layer.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.