# Summary

You can now explain why an AI-native org treats content as one graph rather than
a collection of silos - with Cogitave's estate as the worked example - and how
that graph stays honest as it grows.

In this module, you:

- Learned the thesis, worked through Cogitave's Core architecture spec: docs,
  the IDP, governance, and infra are **projections of one graph**, never
  separate stores.
- Traced how a source document becomes a node: the **ACQUIRE-to-PUBLISH**
  pipeline, the **UID**/**contentHash** identity scheme, and the closed edge
  set (`xref`, `partOf`, `appliesTo`, `teachesSkill`) that links it in - served
  identically to humans and agents through `docs_fetch` and the
  `cogitave://{type}/{id}` resource surface in the MCP interface.
- Learned why linked edges are not the whole propagation story: Cogitave's
  knowledge-propagation standard names **restatement drift** and fixes it
  with a **fact registry** - one owner document per fact, everything else
  cites - checked by a deterministic **fact-drift scanner**.
- Read the status of Cogitave's Core honestly: it is a **specification and
  architecture today**; the registry and scanner already run over files on the
  mirror, and full graph projection lands when Core runs.

## You have completed the path

This was the final module of **Build on a single canonical model**. Completing
it earns the path's trophy - you now have a working model of the one graph every
product, standard, and agent in an AI-native estate shares (Cogitave's own being
the worked example), and how to keep it truthful as content grows.

## Where to go next

- @cogitave.learn.paths.build-on-core - return to the path to claim your
  trophy.
- The **Core architecture** doc, in Cogitave's core repository, is worth a
  re-read for the full node/edge model and the change pipeline at the source.
- The **knowledge-propagation standard** is the canonical reference for the
  fact registry format, mention rules, and the impact-map skill in full.