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

# Oracle connections

> Create a connection and let ORCA discover the pod's metadata

A connection is **one Oracle Fusion pod, its credentials, and its own discovered metadata catalog**. Everything ORCA does runs through a connection, and each pod carries its own catalog — nothing is shared between connections.

## Create a connection

<Steps>
  <Step title="Open the connections tab">
    Go to **Settings → Oracle Connections** and click **Add Connection**.
  </Step>

  <Step title="Fill in the form">
    In **New Connection**, enter a **Name**, the **Oracle URL** (your pod base URL), a **Username**, and a **Password**.
  </Step>

  <Step title="Set options and save">
    Optionally tick **Set as default connection**. Leave **Production connection** unticked for a test pod. Click **Save**.
  </Step>
</Steps>

<Info>
  The moment you save, ORCA starts **background metadata discovery** against the pod — you don't trigger it manually.
</Info>

## Metadata discovery

A **Discovery** row of chips appears on the connection card, one per source. Each starts grey (`never run`), spins while it works (`pending → running`), and turns green (`done`) with an object count — or red (`failed`) if the pod refused. Re-trigger the whole sweep any time with the **Re-run discovery** button.

| Source  | Discovers                                 | Feeds                                                |
| ------- | ----------------------------------------- | ---------------------------------------------------- |
| `rest`  | HCM + FSCM REST resource catalog          | The "Discovered APIs" table and the chat/query tools |
| `hdl`   | HCM Data Loader business objects          | HDL generation for HCM conversions                   |
| `fbdi`  | Finance / SCM / Projects import processes | The "FBDI Processes" table and FBDI loads            |
| `table` | Table / column comments (SQL only)        | `search_tables`; skipped when SQL is off             |

<Note>
  Two more chips (`fsm`, `soap`) render but are placeholders today — they always record as skipped. For HCM work, `rest` and `hdl` are the ones that matter.
</Note>

## What a connection powers

One connection powers **ORCA chat**, the **conversion loads** a cycle submits, and the **MCP connector** that lets Claude query the pod from outside the app. Flip the **MCP** toggle and ORCA surfaces a shared **OAuth URL** (recommended) plus a legacy token URL.

<Card title="Connect Claude to ORCA" icon="robot" href="/orca/integrations/connect-to-claude">
  Query your Oracle data from Claude Desktop or Claude Code over MCP.
</Card>

## Next steps

<Card title="Governed REST vs SQL" icon="shield-halved" href="/orca/connections/governed-rest-vs-sql">
  How ORCA stays inside your Oracle security — and when SQL is allowed.
</Card>