> ## 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.

# Twin: Internal Architecture

> Internal notes on how Twin is composed: identity binding, context intake, reasoning loop, capability invocation, and continuity across sessions.

This page is the internal architecture note for Twin. For the product-level view, see [Product: Twin](/product/twin) and [Contextual Twin](/how-it-works/contextual-twin).

## Composition

* **Identity binding** — Twin runs under the user's One Auth identity.
* **Context intake** — reads assembled context from Spine on focus and on invocation.
* **Reasoning loop** — observe, orient, reason, propose.
* **Capability invocation** — every write goes through governed capabilities.
* **Continuity** — memory of prior work, decisions, and preferences persists.

## Related

* [Core Agents](/internal-docs/agents/core-agents)
* [Spine Agent Network](/internal-docs/agents/spine-agent-network)
* [Product: Twin](/product/twin)


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