Design with AI
SwaggerHub HaloAI is an AI-powered feature that allows you to create, edit, or enhance OpenAPI and AsyncAPI definitions. By describing an API in natural language, you can automatically generate or update a complete API definition, streamlining the design process and reducing the need for manual editing. The APIs generated with HaloAI aim to align with the standardization rules defined in your organization.
Create a new API
To generate a new API with the help of AI features, perform the following steps:
Select the Generate with HaloAI tile visible on SwaggerHub landing page.
Alternatively, click Create New, and select Generate with HaloAI.
Provide the details of the API you want to create by selecting the relevant specification and (optionally) project in the Generate API with HaloAI dialog. To improve accuracy, reference a template to help the AI better understand the type of definition you want to create, and generate a more aligned output.
Enter a prompt describing your API. Based on this natural language description, HaloAI will generate a definition that you can then further refine, if necessary. For best results, provide a detailed and precise prompt.
(Optional) Alternatively, select the try an example option to have HaloAI provide an example prompt you can use as a reference.
Edit an API
With HaloAI, you can also update your existing APIs by describing the desired enhancements in natural language.
Perform the following steps to edit an API:

Open the API in the Code Editor and click the Open HaloAI icon on the toolbar.
Navigate to the Edit and Improve API with AI input available in the HaloAI tab.
Enter a prompt detailing the updates you want to make to your API definition.
After HaloAI applies the requested modifications to your API definition, SwaggerHub revalidates the API against your organization's standardization rules and display an updated list of errors.
Enable or disable HaloAI
As an organization administrator, you can choose whether to enable HaloAI features for your organization. If you disable the feature, all HaloAI components are hidden from the application interface.
Note
This configuration option is accessible in both SaaS and On-Prem.
To configure this setting, perform the following steps:

Go to Organization Settings.
in the section AI Features, toggle Enable AI features on or off.
Run HaloAI on your On-prem instance
You can configure HaloAI availability across your entire On-Prem instance, provided your license includes access to AI features. To request such a license, contact our Sales Team or your account manager.
Depending on the type of SwaggerHub installation, edit this setting in one of the following ways:
For a KOTS installation: Select the Enable SaaS AI features checkbox in the Admin Console. Note that this option is also active by default.
If your license does not cover AI features, the checkbox is disabled.
For a Helm installation: During the installation process, enable access to HaloAI in the pre-prepared custom chart value. Note that this option is set to
true
by default.
If you disable access to AI features across the entire instance, this configuration will also be unavailable at the organization level.
Network configuration
Note
This configuration refers to On-Prem installations only.
To use the optional AI capabilities, your network must allow outbound connectivity to a cloud-hosted SaaS API.
Follow the steps below to add the required setup:
Ensure that the host environment has internet access.
Enable TCP port 443 to allow connectivity over HTTPS to the required SaaS endpoint for AI.
If your network uses a proxy or outbound firewall, whitelist the following domain:
api.swaggerhub.com
.Obtain an updated license key from your Sales Representative, containing your customer-specific API key information to authenticate against the SaaS endpoint for AI.
Note: All AI-related traffic originates from the swaggerhub-frontend
pod and is sent exclusively to this domain over HTTPS. The customer-specific API key is securely stored in the application configuration and used only for AI requests. SwaggerHub does not store any prompts, responses, or parts of the specification data sent from on-premise installations. Only transient request metadata, such as timestamp, API version, installation identifier, etc., may be logged for operational purposes. All API interactions are logged locally in the pod at the INFO level and can be audited via swaggerhub-frontend
pod sysout monitoring.