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

# Collections

> Collections each Instance gets from the Graph.

**Collections** on the Graph toolbar lists the collections an Instance should already have. Save the Graph, then publish so the pinned version carries the list.

<h2 id="edit">
  Edit
</h2>

**Collections** opens the list. **Add collection** appends a row. **Save** writes the list into the Graph draft (autosave).

| Field         | What it does                                                                  |
| ------------- | ----------------------------------------------------------------------------- |
| **Name**      | Letters first, then letters, digits, `.`, `_`, `-`. Example: `carrier_rates`. |
| **Schema**    | JSON Schema for one record. The root is an object.                            |
| **No schema** | Records stay free-form.                                                       |

At most 32 collections. A name already in the list cannot be saved. Schemas that are too large cannot be saved.

Under **Import from Instance**, choose an Instance and press **Import**. That copies names and schemas, not records. A matching name replaces that row. Remove rows you do not want, then **Save**.

Removing a row does not delete records already stored on an Instance.

<h2 id="instance">
  After publish
</h2>

A new Instance lists these collections on [State](/instance/state#workflow), empty, with the badge **Workflow**.

**Save** also updates Instances of this workflow that are not pinned, when stored records fit the new schema. If they do not, the dialog names those collections and they keep the previous schema. The Graph draft is still saved.

Pinning an Instance to a published version applies that version's list. If stored records do not fit, the Instance stays on the current version.