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

# Manage Custom Images

> View custom image details in Siverse, determine whether an image is suitable for reuse, and manage image sharing, cloning, build tasks, and deletion.

After a custom image is built or imported, it appears in the **available images list**. Here, you can verify whether an image is suitable for reuse in subsequent workloads, view its build source, share it with other project groups or users, or remove images that are no longer in use.

## Confirm Image Availability

Go to **Images > Custom**, open the **Available Images** tab, and find the image you want to use.\\

![Custom Image Available Images List](https://mintcdn.com/scitix-adfc65c2/ZqwdmfntBUDyGPkJ/siverse/images/media/custom-image-list.png?fit=max\&auto=format\&n=ZqwdmfntBUDyGPkJ\&q=85\&s=f9d30beb75fcf1bca383f87f8b4ef4be)

Before using an image for a development environment, training task, inference service, or team sharing, we recommend confirming the following information:

| Check item | Description |
| - | - |
| Version and source | The image name, version, and commit ID meet expectations and can be mapped to a code commit or release version. |
| Target cluster | An available image URL has been generated for the target cluster. For cross-cluster use, first [synchronize the image to other clusters](/siverse/images/sync-images-across-clusters.md), and use the URL displayed on the target cluster page. |
| Visible Scope | The image class, owner, shared project groups, region, and project group visibility scope meet the requirements for subsequent use. |
| Runtime validation | Use a new development environment or a small-scale workload to verify that the image can be pulled and started, and confirm that key dependencies, GPU libraries, framework versions, and startup commands work properly. |
| Rollback preparation | Before replacing an image used in production or shared by a team, retain the previous validated version and confirm application and data compatibility. |

An image being pulled and started only indicates that the runtime environment has passed basic checks. Training tasks and inference services must still be validated separately through training steps, output results, or actual requests.

## Manage Custom Images

After locating the target image in the **Available Images** list, you can perform the following actions as needed:

| Action | Description |
| - | - |
| share | Share an image with project groups or other visible scopes. After sharing, authorized users can select the image when creating development environments, training tasks, or inference services. The available sharing scope depends on the image owner, sharing scope, and current permissions. |
| clone | Create a new image build configuration based on an existing available image. This is suitable for reusing the image source, dependencies, and build parameters. |
| view tasks | View build tasks associated with the image. To view build logs or configuration, configure or update the build task settings, stop a build task, or delete one, go to the **Build Task List**. |
| delete | Delete images that are no longer needed. Before deleting an image, confirm the following:<ul><li>Use [View Image Usage Statistics](/siverse/images/view-image-usage-statistics.md) to confirm that no development environments, training tasks, general services, or LLM inference workloads still depend on the image.</li><li>Workloads that require retrying, rebuilding, scaling out, or historical rollback have alternative images available for pulling.</li><li>Your current account has permission to delete the image. The range of images that administrators and standard users can delete may differ.</li><li>You have confirmed the image sharing scope and the deletion confirmation prompt in the console. Deletion may affect platform visibility or the original data in the registry. Refer to the console prompts for the actual impact.</li></ul> |

Before deleting a shared image, we recommend confirming with the relevant users to avoid affecting development environments, training tasks, or inference services that are still using the image. Existing containers can continue running, but this does not mean that the image can still be pulled when creating, restarting, scaling out, or rolling back containers later.

## Frequently Asked Questions

> **What should I do if I cannot find a built image when creating a workload?**
>
> First, confirm that the current region, target cluster, project-group visibility scope, and filter criteria match the image information. When using the image across clusters, also confirm that the image has been synchronized to the target cluster and use the image URL displayed on the target cluster page.

> **What should I do if training retries, service scaling, or historical rollbacks fail after an image is deleted?**
>
> These actions may require pulling the original image version again. Restore a pullable image version, or use a verified alternative image, and confirm application and data compatibility.

## Related Documentation

* [Synchronize Images to Other Clusters](/siverse/images/sync-images-across-clusters.md)
* [View Image Usage Statistics](/siverse/images/view-image-usage-statistics.md)


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.