Extensions

Overview

Extensions are optional features built into VerifyWise. Some connect VerifyWise to tools you already use (Slack, MLflow, Azure AI Foundry, Jira Assets). Others add extra tools inside the app (Model Lifecycle, Risk Import, Dataset Bulk Upload).

All seven extensions ship with VerifyWise, so there is nothing to download or install. Each one stays off until an Admin enables it for the organization.

Finding extensions

There are two ways to open the Extensions page:

• Header: Click the Extensions button (package icon) in the top header. Its tooltip reads "Enable or disable extensions for your organization."
• Start here: Open Start here in the sidebar and click the Extensions card.

The page lists one card per extension. Each card shows the extension name, its category and version (for example "ML Ops · v1.0.0"), and " · Configuration required" when you need to fill in settings before you can enable it. Enabled extensions carry an Enabled badge.

Enabling and configuring an extension

The buttons on each card depend on the extension and its state:

The settings page (reached through Configure or Configure to enable) has a Back to extensions link and, depending on the extension:

• Configuration: A form with the extension's connection settings. Password and token fields are stored encrypted. Leave a secret field blank to keep the existing value.
• Extension panel: Slack, Model Lifecycle and Jira Assets show their own setup panel here once enabled.
• Enable / Disable: Turns the extension on or off.
• Save configuration: Saves changes to the Configuration form after the extension is enabled. You see "Configuration saved." when it works.
• Test connection: Checks the saved settings against the external service. Available for MLflow and Azure AI Foundry. Jira Assets has its own Test Connection button in its panel.

If something goes wrong, the page shows the error returned by the server, or a short message such as "Enable failed", "Save failed" or "Test failed".

Disabling an extension

Click Disable on the card or on the settings page. The extension turns off immediately, without a confirmation step, and its tabs, buttons and menu items disappear from the app.

Available extensions

Slack

Sends VerifyWise notifications to channels in your Slack workspace.

1. Click Enable on the Slack card, then Configure.
2. Click Add to Slack. Slack asks you to pick a workspace and the channel to post to, and to approve the connection.
3. Back in VerifyWise, the connection appears in the table (Team Name, Channel, Creation Date, Active, Action). You can add more connections the same way.
4. Click Configure to open Notification Routing, choose the notification types under "Apply to all workspaces", and click Save Changes.

Where it shows up: notifications arrive in the Slack channels you connected. See Slack integration for routing, troubleshooting and removing connections.

MLflow

Pulls models and runs from your MLflow tracking server into VerifyWise.

1. Click Configure to enable on the MLflow card.
2. Fill in the Configuration form (see the table below).
3. Click Enable, then Test connection to check the server is reachable.

Where it shows up: an MLFlow tab in Model inventory. Click Sync there to pull the latest runs. Syncing is manual; nothing syncs on a schedule.

Azure AI Foundry

Imports the model deployments in an Azure AI Foundry project. Agent discovery also uses this connection as one of its sources.

1. Click Configure to enable on the Azure AI Foundry card.
2. Fill in the Configuration form (see the table below).
3. Click Enable, then Test connection.

Where it shows up: an Azure AI Foundry tab in Model inventory. Click Sync there to refresh the deployments.

Model Lifecycle

Lets you define lifecycle phases for your models, each with its own items such as approvals, documents and responsible people.

1. Click Enable on the Model Lifecycle card, then Configure.
2. Click Configure Phases to open Configure Model Lifecycle.
3. Click Add new phase, give it a name and an optional description.
4. Inside a phase, click Add item, name it and pick its type: Text, Text Area, Documents, People, Classification, Checklist or Approval.

Where it shows up: a View lifecycle button on each row in Model inventory, and a lifecycle section on each model's detail page.

Risk Import

Creates many risks at once from a filled-in Excel template. Click Enable on the card; there is nothing to configure.

Where it shows up: in Risk management, click Add new risk and choose Import from Excel. The Import Risks from Excel dialog walks you through three steps:

1. Step 1: click Download Template.
2. Step 2: fill in the template and upload it with Choose File.
3. Step 3: click Import Risks.

Columns marked with * in the template are required: Risk Name, Risk Owner, AI Lifecycle Phase, Likelihood and Severity. Each row also needs a Risk Description, or the import rejects that row. Admins and Editors can import risks.

Jira Assets Integration

Imports "AI System" objects from Jira Service Management Assets and turns them into VerifyWise use cases.

1. Click Enable on the Jira Assets Integration card, then Configure.
2. Choose a Deployment Type: JIRA Cloud (Atlassian-hosted) or JIRA Data Center / Server (Self-hosted).
3. Fill in JIRA Base URL, then Workspace ID (Cloud) or Insight Object Schema ID (Data Center), then Email and API Token (Cloud) or Username and Password / Token (Data Center).
4. Click Test Connection, then Save Configuration.
5. Pick the Schema and the Object Type (AI Systems) to import from.
6. In the Import & Sync section, click Import, select the objects you want and click Import N Selected.

Where it shows up: imported objects appear as use cases under Use cases. To refresh them later, click Sync Now in the Import & Sync section.

Dataset Bulk Upload

Registers many datasets at once. Click Enable on the card; there is nothing to configure.

Where it shows up: a Bulk upload button on the Datasets page, available to Admins and Editors. You can upload up to 20 CSV, XLSX or XLS files at a time, up to 30 MB each. VerifyWise checks column headers for likely personal data and flags matching datasets. Developers can use the same upload through the API.

Who can do what