Use Connect Anything when no ready-made source type fits your system. You build a connector from a template, install it, write its configuration and secrets, then load its tables into the Lake. For an HTTP API you can describe as a recipe, use Connect any API instead.
Creating the connector and changing its code, configuration or secrets needs Add/Modify Lake Objects. Starting a sync needs Execute/Abort Lake Objects. With only View Lake Objects, you can see the connector and its configuration but cannot change them. See Who can do what in the Lake.
The wizard has four steps: General, Build, Connection and Review.
After creation, Name and Code are locked. You can still change the Folder.
The Build step shows the install card and the connector template card. The install runs through seven steps: Create the project, Add the connector, Install the connector, Add the loader, Install the loader, Configure logging and Done, headed Step {n} of 7. Expand a step for its detail; Copy copies the detail and shows a check for two seconds.
The install card has one of four states:
| State | What it means | What to do |
|---|---|---|
Installing… | The install is running. Syncing waits for it. | Wait. It takes about a minute. |
Install failed | The install did not finish. The error stays on the step. | Click Retry. |
Reinstall needed | The code changed since the last install. | Click Reinstall. |
Installed | The connector is ready to sync. | Continue to Connection. |
Loading the connector…, Wait for the install to finish, Fix the install error and retry or Reinstall the connector to continue.Live progress is unavailable - checking the status every few seconds. and keeps checking the status.Code changed since the last install and Reinstall to apply it - new dependencies from pyproject. Reinstall installs the new code and its new dependencies. Until then the connector cannot sync, and after a successful reinstall it syncs again.To change the code, click Open code editor on the connector template card and work with the connector agent there, or edit the files yourself. The connector agent can also read and edit the files, read and validate the configuration, reinstall the connector and run a load to prove the data arrives; any file change still means a reinstall before the next sync. See The connector code editor. Until the change is reinstalled, the connector cannot sync: "Not synchronized: the connector has code changes that are not installed. Reinstall it first."
If you created the connector by mistake, click Delete connector on the install card while you are still in the create flow. The confirmation says it "permanently deletes the connector you just created and its code. This cannot be undone."
The Connection step holds the Configuration (meltano. editor and the Secrets section. While the install is not finished, the step shows a notice with Go to Build. If a load failed earlier, the step shows Last sync failed.
AI Configuration Suggestion) with Reject and Accept; Enter accepts. Secret values it hides stay masked until you deal with them.{{secret:NAME}} reference in place of the value. A stored value is never shown again; use Replace to change it. An inline value in the file can be moved with Move to Secrets.Add the configuration before continuing.The vault's full rules are on Connector secrets; the editor's detail is on Configure a custom connector.
On Review, check the summary rows: Name, Code, Folder, Connector, Configuration and Secrets.
Start the first data load right after Add.The first load is a capped sample by design: every stream stops after 100 records. Later runs load fully, and scheduled runs are not capped. A stream that lands fewer rows than the cap is complete, not cut short.
Open the connector from the Lake tree and choose Edit. Name and Code stay locked.
To remove a connector you no longer need, delete it from the Lake tree. See Delete Lake items. A chat with a Lake Builder Mind can also manage connectors for you; see Manage connectors from a chat.
About a minute. The Build card shows the live progress with Step {n} of 7 and keeps the finished state after the install ends.
The connector must be installed first. The hint says what it waits for: Loading the connector…, Wait for the install to finish, Fix the install error and retry or Reinstall the connector to continue.
Yes. Click Reinstall on the Build step, or use the same action in the code editor. Until the reinstall finishes, the connector cannot sync: "Not synchronized: the connector has code changes that are not installed. Reinstall it first." Reinstalling also discards the connector's incremental bookmarks and run history.
Save stores your changes without reloading data. Save and sync stores them and starts a full data load. A load cannot be cancelled once it starts, so use Save and sync only when you want the data refreshed now.
By design. The connector's first load stops after a sample of 100 records per stream; later runs load fully, and scheduled runs are not capped. A stream with fewer rows than the cap is complete.
Store the value in the Secrets Vault and paste its {{secret:NAME}} reference into the configuration in place of the value; the form's own steps are on Configure a custom connector.
Yes, through the configuration's History; The connector code editor covers the list and how to restore one.
It writes the configuration from your description and puts the result into the editor as an unsaved suggestion. Inline secret values it finds are hidden, so move each one to the vault before you test.
If you are still in the create flow, click Delete connector on the install card and confirm. The message says it "permanently deletes the connector you just created and its code". After a successful install, delete it from the Lake tree; see Delete Lake items.