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

# Core Concepts

> Current Arcyria concepts: connections, the pipeline workspace, source streams, destination-owned transformations, output assignments, runs, and schedules.

## Connections

Connections store reusable encrypted credentials. PostgreSQL, MySQL, MongoDB,
and Airtable can be sources or destinations. MongoDB stores its complete URI
as one encrypted secret and is currently Available Now. Asana, Notion, Strapi, HubSpot,
and Stripe are sources. Strapi is an Available Now, Strapi 5-only source with dynamic
Content API IDs and Full Table sync.

Source connections need read access. Destination connections need write access
to the configured tables or collections.

## Pipeline workspace

A pipeline starts with a name and source connection. Opening it shows the
current tabbed workspace: **Overview**, **Source**, **Transformations**,
**Destinations**, **Runs**, **GitHub**, and **Settings**.

The old canvas and Source/Destination nodes no longer exist.

## Source streams

Use the **Source** tab to:

* test the source connection;
* discover or refresh the catalog;
* select one or more streams;
* choose Full table or Incremental per supported stream;
* provide an Incremental cursor;
* inspect fields and primary keys; and
* preview representative rows.

Streams use stable keys such as `public.orders`, `application.orders`,
`stripe.customers`, `airtable.tbl...`, or `notion.data_source_...`. Database staging relations use
names such as `public__orders`; Notion data-source relations use
`notion__data_source_...`; Strapi collections use `strapi__<api_id>`.

## Destinations

Use **Destinations** to add an independent delivery target. Each destination has
its own connection, display name, schema/database or Airtable base/table, SQL
transformations, output assignments, Upsert keys, and checkpoint behavior.

Adding a destination opens a focused destination editor; it is not a node on a
canvas.

## Transformations and revisions

Transformations are owned by one destination. Use **Transformations** to select
a destination and create its SQL model. The release sequence is:

1. **Save draft**
2. **Validate draft**
3. **Preview output**
4. **Publish revision**

Draft changes do not affect runs until published. Each destination can evolve
without changing another destination's published revision.

Rename, cast, filter, join, derive, and aggregate values directly in SQL. The
current workspace does not expose the former Normalisation panel.

## Destination tables and Upsert

Published model output is assigned to an exact destination table or collection.
Database targets use `schema.table` or `database.table`; MongoDB targets use
`database.collection`; Airtable targets use the
selected existing table and explicit writable-field mappings.

The current delivery mode is **Upsert**. Configure stable, non-empty Upsert keys
that match a primary/unique database key, indexed MongoDB field, or writable
Airtable merge fields.

## Overview and readiness

The **Overview** tab summarizes the shared source, pipeline settings,
destination-owned transformations, and delivery targets. **Validate pipeline**
checks that source streams, published revisions, and destination assignments
are ready before **Run all** is enabled.

## Runs and schedules

Use **Runs** for statuses, destination targets, published revision IDs, row
counts, durations, warnings, and sanitized errors.

Use **Settings** for activation, validation, and scheduling. The current
schedule editor supports a fixed 1–24 hour interval or a five-field UTC cron
expression. Validate a manual run before enabling automation.