> AI agents: this is one page from Mammoth Analytics documentation. The index of all pages as Markdown is https://mammoth.io/docs/llms.txt. Append `.md` to any docs URL, or send `Accept: text/markdown`, to get Markdown.

# Manage saved connections on the Connections page

> See every saved connection in a project, check its health and usage, start a dataset from it, edit or delete it, and track datasets that follow a link.

A connection is a saved sign-in to a source or destination: a database login, an app you signed in to, or an API key. You create one the first time you connect a source in the [connector workspace](https://mammoth.io/docs/learn/connectors/connector-workspace/), or when you set up an export. After that, the connection belongs to the project, and the **Connections** page lists it with everything that uses it.

## Open the Connections page

Go to **Data** and click the **Connections** tab (beside **Datasets**). The page lists every connection in the current project:

- **Mammoth Managed**: connections Mammoth runs for you, such as a managed PostgreSQL or Google BigQuery database you export to. See [Send to destinations](https://mammoth.io/docs/learn/export-share/send-to-destinations/).
- **External**: connections you or a teammate created, to your own databases, apps, drives and APIs.

Each row shows the connection's name, its connector, a health dot, and how it is used: a source count (datasets that import from it) and a destination count (views that export to it). Hover over a count to see the datasets or views by name, and click one to open it. A connection nobody uses shows **No linked datasets**. One with a dataset still being set up shows how many, such as **1 in setup**.



## Who can see and change connections

Connections are shared across the project. Everyone who can open the project's **Data** area sees the same list. Passwords and keys of your own connections are never shown on the page.

| Action | Who can do it |
| --- | --- |
| See the list, a connection's details and its usage | Everyone in the project |
| Start a dataset from an existing connection (**Create dataset**) | Project Admins and Project Editors |
| Create, edit or delete a connection, discard a draft | Project Admins |
| See a Mammoth-managed database's credentials | Project Admins |

If you can't change connections, the page shows a banner and the buttons are disabled:

- "You don't have permission to create or change connections in this project. Ask a Project Admin to add a connection or to give you Admin access."
- In a workspace without an active plan: "This workspace is in view-only mode, so connections can't be created or changed. Choose a plan to continue working."

See [Project roles and permissions](https://mammoth.io/docs/learn/projects/project-roles-and-permissions/) for the roles.

## Search and filter connections

Type in **Search connections** to match a connection's name or its connector (for example, `postgres`).

The tabs above the list filter it. Each tab shows a count.

| Tab | Shows |
| --- | --- |
| **All** | Every connection, plus unfinished drafts and the **Links** group |
| **Mammoth** | Mammoth-managed connections |
| **External** | Connections you or a teammate created |
| **Sources** | Connections at least one dataset imports from, plus drafts and **Links** |
| **Destinations** | Connections at least one view exports to |
| **Issues** | Connections whose health is **Error** |
| **Unused** | Connections nothing imports from or exports to, and with no dataset in setup |

**All**, **Sources** and **Destinations** always show. The other tabs appear only when they have something in them. **Mammoth** and **External** are hidden when every connection is external.

When connections or links are failing, a line above the list shows how many "need attention".

If the search or a tab matches nothing, you see "No connections match your search." with **Clear search** or **Clear all filters**.

## Connection details

Click a connection to open its details beside the list.

The top shows the connection's health and, depending on its state, when it was last checked, **Not in use**, or how many datasets are **In setup**.

### Health status

| Status | What it means |
| --- | --- |
| **Connected** | The connection's last read worked |
| **Error** | The key expired, the source refused it, or Mammoth could not reach the source |
| **Unknown** | Mammoth has not checked this connection yet |

A connection's health comes from a daily check and from real reads: a table that returns rows in the connector workspace marks it **Connected**, and a source that refuses the key marks it as failing. A failing export to the connection also marks it **Error**.

### Connection Details

Under **Connection Details** you see the **Connector**, **Managed by** (**Mammoth Managed** or **External**), **Health status**, **Auth method**, and, where the connection has them, the **Auth fields**, **Host**, **Port**, **Database**, **Username** and **Encryption**. **Added by** and **Added** say who created it and when.

### Usage

**Usage** lists the **SOURCES** (datasets that import through the connection) and the **DESTINATIONS** (views that export to it), each with its folder path. Click one to open it. A connection that nothing uses says "This connection is not used by any dataset yet."

### Actions

| Button | What it does |
| --- | --- |
| **Create dataset** | Opens a new draft on this connection in the connector workspace, with nothing to re-enter. |
| **Continue** and the draft's name, such as **Continue HubSpot deals** | Shown when one dataset from this connection is still in setup: reopens that draft. **Create dataset** then reads **New dataset**. |
| **Edit connection** | Opens the connection form to change its details, such as a new password, key or sign-in. Built-in connectors only. |
| Delete (bin icon) | Deletes the connection. See [Delete a connection](#delete-a-connection). |
| **Show credentials** | Mammoth-managed PostgreSQL and Google BigQuery only. See [Mammoth-managed connections](#mammoth-managed-connections). |

When you save an edit, Mammoth shows, for example, "Connection Sales Postgres updated".

## Fix a connection that needs attention

The Connectors launcher lists failing connections under **Needs attention**, with a reason, **key expired** or **unreachable**. Click one: the Connections page opens with that connection selected, and:

- For a built-in connector, the **Edit connection** form opens so you can enter a new key or sign in again.
- For a connection the agent built (a REST API), a new draft opens and its chat asks for the key again.
- For a Mammoth-managed connection: "Mammoth manages this connection's credentials, so there is nothing to repair here."
- If you are not a Project Admin: "Only a project admin can repair this connection."

## Delete a connection

1. Open the connection and click the delete (bin) icon.
2. Read the dialog, titled with the connection's name, such as **Delete connection “Sales Postgres”?**. It lists what stops working:
   - **Imports**: each dataset that "Stops refreshing. Rows already imported are kept."
   - **Exports**: each task that "Stops delivering. Rows already delivered stay in the destination."
   - **Stored credentials**: "You’d re-enter these to reconnect".
3. Click **Delete connection**.

Deleting a connection never deletes data, datasets or pipelines. Its credentials are removed, and everything that used it becomes **unconfigured** until you point it at another connection. Mammoth confirms, for example, "Connection Sales Postgres deleted. 3 usages are now unconfigured."

If nothing uses the connection, the dialog says "Nothing is using this connection." A deleted connection does not go to the trash, so you can't restore it: create it again instead.

Mammoth-managed connections have no delete button. Mammoth removes one by itself when the last published view that uses it is deleted.

## Mammoth-managed connections

Mammoth creates a managed connection when you export a view to a Mammoth-managed PostgreSQL or Google BigQuery database. To connect an outside tool (for example, a BI tool) to that database:

1. Open the managed connection and click **Show credentials**.
2. The **Database credentials** dialog lists the values: for PostgreSQL, the host, port, username, password and database; for Google BigQuery, the dataset, the service account's email and a link to the service account key.
3. Click **Copy all** to copy them.

Only Project Admins see **Show credentials**.

## Add a new connection

Click **New Connection**:

- On any tab except **Destinations**, it opens the [Connectors launcher](https://mammoth.io/docs/learn/connectors/connector-workspace/#open-the-connectors-launcher), where you pick a source and connect it in the connector workspace.
- On **Destinations**, it opens a picker of connectors. Select one, click **Next: Configure**, and fill in the connection form (or sign in). When it's saved, **Connection Created** offers **Create a dataset** or **Go to Connections**.

A project with no connections yet shows "No connections yet" and a **New Connection** button.

## Drafts on the Connections page

A source you started in the connector workspace but didn't save is a draft. On **All** and **Sources**:

- Drafts that are not connected yet are listed at the top under **Not connected yet**. A draft whose key the chat already checked shows **Key checked**.
- A draft on a saved connection sits under that connection's row.

Click **Continue** to reopen a draft. Project Admins can click **Discard**: **Discard this draft?** warns "The draft and its chat go away. Nothing in your data changes."

## Links: datasets that follow a link

The **Links** group, under the connections on **All** and **Sources**, lists every dataset created from a [file at a URL](https://mammoth.io/docs/learn/data-sources/import-from-url/), grouped by website. Each website row shows how many datasets follow it, how many are on a schedule, and **Last fetch failed** if one failed.

Open a website to see its datasets. Each shows when it was last fetched (such as **Fetched 2 hours ago**) or why the last fetch failed (for example, "The link answered 404" or "The link could not be reached"), and its schedule or **Manual**.

| Option | What it does |
| --- | --- |
| **Get the latest** | Fetches the file again now |
| **Automate the refresh** (menu) | Sets up a scheduled refresh for the dataset |
| **Open dataset** (menu) | Opens the dataset |

When two datasets follow the same link, the second says, for example, **same link as Daily Sales**, with a note that "Usually one was added by mistake."

## Related pages

- [Add a connector with the connector workspace](https://mammoth.io/docs/learn/connectors/connector-workspace/)
- [Connectors](https://mammoth.io/docs/learn/connectors/)
- [Import from URL](https://mammoth.io/docs/learn/data-sources/import-from-url/)
- [Send to destinations](https://mammoth.io/docs/learn/export-share/send-to-destinations/)
- [Connection troubleshooting](https://mammoth.io/docs/learn/troubleshooting/connection-troubleshooting/)

---
Source: https://mammoth.io/docs/learn/connectors/connections.md · Updated: 2026-11-03