> For the complete documentation index, see [llms.txt](https://docs.kinesis.network/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.kinesis.network/getting-started/running-llms-with-kinesis-network/creating-an-app-from-scratch.md).

# Create a custom app

Configure a container image or GitHub repository in the five-step app wizard.

Select the intended organization, open a project you can edit and select **Create App**. You need an Admin or Developer role with access to that project. The wizard groups configuration into five steps.

| Step          | Configure                                                                            | Check before continuing                                                    |
| ------------- | ------------------------------------------------------------------------------------ | -------------------------------------------------------------------------- |
| **Source**    | Container image or GitHub repository                                                 | Image tag and registry access; or branch, Dockerfile and context directory |
| **Runtime**   | Environment variables, commands, health checks, shared memory and persistent storage | Required secrets and the process the image should run                      |
| **Network**   | Exposed ports, allowed IPs and domains                                               | Container port, protocol and listener binding                              |
| **Execution** | Service or task execution and restart behavior                                       | Whether the workload should stay running or finish                         |
| **Resources** | Serverless requirements or grids, placement and scale                                | GPU memory, RAM, storage, instance bounds and capacity                     |

## Container images

Use an image reference the runtime can pull. Supply credentials for a private registry. Environment variables marked as secrets are write-only: API reads do not reveal their values.

Override the image's command or entrypoint only when the workload requires it. A service that exits immediately may have the wrong startup command even when its image builds successfully.

## GitHub repositories

Connect the Kinesis GitHub App to the account or organization containing the repository, then select the repository and branch. Confirm the Dockerfile and build context. The portal can use an inline Dockerfile override when needed.

Repository deployments go through import, image build and deployment. Open **Deployments** to follow progress and inspect builder logs. Automatic deployment can rebuild on pushes to the configured branch when enabled.

The public API expresses this source as `source: { kind: "github", installationId, repo, ... }`. The build supplies `config.image`, so do not set a separate image for a repository app.

## Roll back a repository deployment

Open **Deployments** and choose an earlier build with a stored image. **Roll back** reuses that image without rebuilding it; it is unavailable when the image is gone. A running app redeploys; a stopped app uses the image on its next start. Wait for an active build to finish first. The next build, including an automatic build after a push, replaces the rollback. See [Builds API](/api/builds.md).

## Network and readiness

The configured port is the port inside the container. Public listener ports depend on the selected protocol: fixed HTTP/HTTPS listeners use 80/443, while other load-balanced protocols may allocate a public port. Use the returned endpoint instead of assuming they are the same.

Bind the service to `0.0.0.0`. A service listening only on localhost cannot receive the app's incoming traffic.

## Resources and persistence

GPUs run on grids only. GPU controls are available for Grid apps; serverless placement is CPU-only. An existing serverless app with a GPU requirement must move to a grid or drop that requirement before its next configuration save.

Use the model or application's actual resource requirements to choose memory and GPU capacity. Shared memory is a separate setting from RAM. Named persistent volumes require grid-backed apps; check the volume mount paths and declared sizes.

Review the app creator’s grid quota and available capacity before launch. A quota refusal may require lowering minimum instances, stopping another of the creator’s apps or asking an organization admin to adjust the policy. If serverless capacity is disabled for Developers, select an accessible grid.

Next: [Compute options](/getting-started/running-llms-with-kinesis-network/understanding-compute-sources.md), [Monitor and troubleshoot](/getting-started/running-llms-with-kinesis-network/monitoring-apps.md), or the [Apps API](/api/apps.md).


---

# Agent Instructions
This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com.

## Querying This Documentation
If you need additional information that is not directly available in this page, you can query the documentation by asking a question.

Perform an HTTP GET request on the following URL with the `ask` and `goal` query parameters:

```
GET https://docs.kinesis.network/getting-started/running-llms-with-kinesis-network/creating-an-app-from-scratch.md?ask=<question>&goal=<user_goal>
```

`ask` is the immediate question: it should be specific, self-contained, and written in natural language.
`goal` is what the user is ultimately trying to achieve, the reason they need the answer. Sharing it helps GitBook give you a better, more relevant answer. A goal is most helpful when it describes the outcome the user wants rather than restating the question. For example, with `ask=how do I create an API token`, a goal like `automate deployments from our CI pipeline` lets GitBook tailor the answer to that use case.

The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.