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

# Integrations

> See the AI apps your people sign in to and install, from your identity provider and device management.

See which AI apps your people actually use, even the ones nobody set up. Identity providers show the apps people sign in to or give access to work data; device management shows the apps installed on managed devices.

For example, connect Microsoft Intune, and an AI desktop app installed on 12 managed laptops shows up in [Discovery](https://docs.xenovia.io/discovery/overview.md) with the devices it is on and the people who use them. If nothing else can govern it, Xenovia can have Intune uninstall it.

These integrations are detection sources: they show that an app is used, but can't govern what happens inside it. Each one reads only, and changes nothing unless you apply a fix that blocks or removes an app.

| Integration | Kind | Can block or remove an app |
| - | - | - |
| [Okta](#okta) | Identity provider | Yes, with `okta.apps.manage` |
| [Microsoft Entra ID](#microsoft-entra-id) | Identity provider | Yes, with `Application.ReadWrite.All` |
| [Google Workspace](#google-workspace) | Identity provider | No |
| [Jamf](#jamf) | Device management | Yes, with Create Restricted Software |
| [Microsoft Intune](#microsoft-intune) | Device management | Only apps Intune deploys, with `DeviceManagementApps.ReadWrite.All` |

You need to be an admin of your Xenovia organization to connect any of them. Each keeps its secret encrypted and never shows it again, and uses it only to get short-lived tokens. Choose **Sync now** on an integration's page to sync right away; **Recent syncs** shows how the last ones went. To change its details, choose **Settings** on the integration page: a field you leave empty keeps its saved value, and Xenovia checks the new values before saving them.

## Okta

![Okta](https://mintcdn.com/xenovia/RO5UF1Tk8L9B7yPt/images/logos/okta.svg?fit=max\&auto=format\&n=RO5UF1Tk8L9B7yPt\&q=85\&s=602b5baa0ee5770740c1b7445fea7f30)

![Okta](https://mintcdn.com/xenovia/RO5UF1Tk8L9B7yPt/images/logos/okta-dark.svg?fit=max\&auto=format\&n=RO5UF1Tk8L9B7yPt\&q=85\&s=aaf2a4390f872ec42335e04b5aef0a98)

The Okta integration reads your Okta apps, who is assigned to them, and your single sign-on events. Each AI app your people use shows up in [Discovery](https://docs.xenovia.io/discovery/overview.md) with the people who use it and the scopes it was granted. Okta is a detection source: it shows that an app is used, but can't govern what happens inside it.

When nothing else can govern an app, Xenovia can block it in Okta for you. Blocking is optional and needs an extra permission.

### Before you start

In Okta, create one of these:

* **An API token** from a read-only admin. Xenovia sends it as an `SSWS` token.
* **An API service app** that uses a public and private key pair. Grant it the scopes `okta.apps.read`, `okta.logs.read` and `okta.appGrants.read`.

`okta.appGrants.read` lets Xenovia see the scopes each AI app was granted. To let Xenovia block an app, also grant `okta.apps.manage`.

### Connect Okta

**1. Open the integration**

In **Settings › Integrations**, find **Okta** under **Identity providers** and choose **Connect**.

**2. Enter your details**

| Field | What to enter |
| - | - |
| **Okta domain** | Your Okta address without a path, for example `acme.okta.com` |
| **API token** | The token. Leave it empty to use a service app instead |
| **Service app client id** | The service app's client id, when you don't use a token |
| **Key id (kid)** | The key id of the service app's key, if Okta gave one |
| **Service app private key** | The PEM private key that matches the service app's public key |

Give either the API token, or the client id and private key.

**3. Connect and check**

Choose **Connect**. Xenovia checks that it can read your apps and your system log. If a check fails, Xenovia shows why and saves nothing.

Xenovia keeps the token and private key encrypted and never shows them again. With a service app, the private key stays with Xenovia: it is used only to get short-lived access tokens from Okta.

### What Xenovia reads

| What | How often |
| - | - |
| Your active apps and the people assigned to them | Every 6 hours |
| Sign-ins through single sign-on, counted per app and person | Every 30 minutes |

The first sync reads the last 30 days of sign-ins. Later syncs continue from where the last one stopped. Xenovia keeps only the apps its AI app catalog recognizes and skips the rest. It only reads; it changes nothing in Okta unless you ask it to block an app.

### Block an app

Open the app in Discovery and choose **Plan a fix**. When nothing else can govern the app, the plan can propose **Block it in Okta**. Applying it deactivates the application in Okta, so nobody can sign in to it. To undo it, activate the application again in Okta.

Xenovia only blocks apps that Okta itself reported for that entry. If Okta lists the app more than once, choose which **Okta app id** to block.

### Disconnect

Choose **Disconnect** on the Okta integration page. Xenovia stops syncing and deletes the stored credentials. Open findings from Okta are resolved, and apps only Okta reported leave Discovery. Your reviews of those apps are kept. To cut off access on Okta's side too, revoke the token or deactivate the service app in Okta.

> [accordion] Troubleshooting
>
> | Message | What to do |
> | - | - |
> | Okta domain must be a public https\:// address | Enter your Okta domain only, such as `acme.okta.com`, with no path, port or IP address |
> | Give an API token, or a client id and a private key | Fill in one of the two ways to connect |
> | Okta rejected the credentials | The token or service app isn't valid. Create a new one |
> | Okta refused access: the credentials need okta.apps.read (or okta.logs.read) | Grant the scope named in the message |
> | Okta refused the service app's token request (check the scopes granted to the app) | Grant the service app all three read scopes |
> | The private key is not a PEM RSA private key | Paste the full PEM private key, including its first and last lines |
> | Okta is rate limiting this tenant; try again shortly | Wait, then choose **Check again** |
> | Okta refused: the credentials need okta.apps.manage | Blocking needs `okta.apps.manage`. Grant it, then apply the fix again |
> | Okta has no app with that id | The app was removed from Okta. Sync again |

## Microsoft Entra ID

![Microsoft Entra ID](https://mintcdn.com/xenovia/RO5UF1Tk8L9B7yPt/images/logos/entra.svg?fit=max\&auto=format\&n=RO5UF1Tk8L9B7yPt\&q=85\&s=ece6b83c019968b93ab84ca8be5c0fac)

The Microsoft Entra ID integration reads your enterprise applications, the permissions people and apps have granted them, and your sign-ins, through Microsoft Graph. Each AI app your people use shows up in [Discovery](https://docs.xenovia.io/discovery/overview.md) with the people who use it and the access it was given. Entra ID is a detection source: it shows that an app is used, but can't govern what happens inside it.

When nothing else can govern an app, Xenovia can block it in Entra ID for you. Blocking is optional and needs an extra permission.

### Before you start

In Entra, register an application and give it these Microsoft Graph **application** permissions, with admin consent:

| Permission | Why |
| - | - |
| `Application.Read.All` | List enterprise applications |
| `AuditLog.Read.All` | Read sign-ins |
| `DelegatedPermissionGrant.Read.All` | See what access people granted each app |
| `User.Read.All` | Name the person behind each grant. Without it, those grants are left out |
| `Application.ReadWrite.All` | Optional: lets Xenovia block an app |

Then create a client secret for the application.

### Connect Entra ID

**1. Open the integration**

In **Settings › Integrations**, find **Microsoft Entra ID** under **Identity providers** and choose **Connect**.

**2. Enter your details**

| Field | What to enter |
| - | - |
| **Tenant id** | The directory (tenant) id, or its domain |
| **Application (client) id** | The registered application's client id |
| **Client secret** | The secret you created |

**3. Connect and check**

Choose **Connect**. Xenovia gets a token for the application and checks that it can list service principals, sign-ins and permission grants. If a check fails, Xenovia shows why and saves nothing.

Xenovia keeps the client secret encrypted and never shows it again. It uses the secret only to get short-lived Graph tokens.

### What Xenovia reads

| What | How often |
| - | - |
| Enterprise applications, the delegated permissions people granted them, and the application permissions they hold | Every 6 hours |
| Successful sign-ins, counted per app and person | Every 30 minutes |

The first sync reads the last 30 days of sign-ins. Later syncs continue from where the last one stopped. Xenovia keeps only the apps its AI app catalog recognizes and skips the rest. It only reads; it changes nothing in Entra unless you ask it to block an app.

### Block an app

Open the app in Discovery and choose **Plan a fix**. When nothing else can govern the app, the plan can propose **Block it in Entra**. Applying it turns off sign-in for the enterprise application, so nobody can sign in to it. To undo it, enable sign-in for the application again in Entra.

Xenovia only blocks service principals that Entra itself reported for that entry. If Entra lists the app more than once, choose which **Service principal id** to block.

### Disconnect

Choose **Disconnect** on the integration page. Xenovia stops syncing and deletes the stored client secret. Open findings from Entra ID are resolved, and apps only Entra ID reported leave Discovery. Your reviews of those apps are kept. To cut off access on Microsoft's side too, delete the client secret or the app registration in Entra.

> [accordion] Troubleshooting
>
> | Message | What to do |
> | - | - |
> | Tenant id must be the directory id or its domain | Enter the tenant id, or a domain such as `acme.onmicrosoft.com` |
> | Microsoft rejected the tenant id, client id or client secret | Check all three, and that the client secret hasn't expired |
> | Microsoft Entra ID refused access: the credentials need Application.Read.All (or another permission) | Add the permission named in the message as an application permission, and grant admin consent |
> | Microsoft Entra ID is rate limiting this tenant; try again shortly | Wait, then choose **Check again** |
> | Entra refused: the app needs Application.ReadWrite.All | Blocking needs `Application.ReadWrite.All`. Add it with admin consent, then apply the fix again |
> | Entra has no service principal with that id | The application was removed from Entra. Sync again |

## Google Workspace

![Google Workspace](https://mintcdn.com/xenovia/RO5UF1Tk8L9B7yPt/images/logos/google_workspace.svg?fit=max\&auto=format\&n=RO5UF1Tk8L9B7yPt\&q=85\&s=dfdb2284cf86fe5b300aaf6c7a1a2f39)

The Google Workspace integration reads your audit reports and your app access settings. Each AI app your people sign in to with Google, or give access to their Google data, shows up in [Discovery](https://docs.xenovia.io/discovery/overview.md) with the people who use it and the scopes it was granted. Google Workspace is a detection source: it shows that an app is used, but can't govern what happens inside it.

Xenovia also reads how you configured each app in the Admin console's API controls, and flags AI apps you haven't configured there.

Xenovia can't block an app through Google Workspace. Google's app access settings can be read but not changed through its API. To stop an app, change its access in the Admin console, or plan a fix in Discovery that uses another integration.

### Before you start

You need to be an admin of your Xenovia organization, and a Google Workspace super administrator to set up access.

**1. Create a service account**

In Google Cloud, create a service account and a JSON key for it.

**2. Delegate it**

In the Google Admin console, open **Security › API controls › Domain-wide delegation**. Add the service account's client id with these three scopes:

```text theme={null}
https://www.googleapis.com/auth/admin.directory.user.readonly
https://www.googleapis.com/auth/admin.reports.audit.readonly
https://www.googleapis.com/auth/cloud-identity.policies.readonly
```

Google refuses the whole token if any one of them is missing.

Google's audit reports are read as a person, not as the service account. Xenovia acts as the super administrator whose address you enter.

### Connect Google Workspace

**1. Open the integration**

In **Settings › Integrations**, find **Google Workspace** under **Identity providers** and choose **Connect**.

**2. Enter your details**

| Field | What to enter |
| - | - |
| **Administrator address** | A super administrator's email address, for example `admin@acme.com` |
| **Service account address** | The `client_email` from the service account's JSON key |
| **Service account private key** | The `private_key` from the same JSON key, starting with `-----BEGIN PRIVATE KEY-----` |

**3. Connect and check**

Choose **Connect**. Xenovia checks that it can read your users, your sign-in reports and your app access policies. If a check fails, Xenovia shows why and saves nothing.

Xenovia keeps the private key encrypted and never shows it again. It uses the key only to get short-lived access tokens from Google.

### What Xenovia reads

| What | How often |
| - | - |
| Third-party apps people authorized with their Google account, with the scopes each was granted, and your API controls settings for them | Every 6 hours |
| Successful sign-ins, counted per app and person | Every 30 minutes |

The first sync reads the last 30 days. Later syncs continue from where the last one stopped. Google's reports can lag by hours, so a sign-in can take that long to appear. Xenovia keeps only the apps its AI app catalog recognizes and skips the rest. It only reads.

When your API controls let unconfigured third-party apps through, an AI app you haven't configured there gets a finding that says whether it can sign people in or reach your Workspace data. Where unconfigured apps are blocked by default, there is nothing to report.

### Disconnect

Choose **Disconnect** on the integration page. Xenovia stops syncing and deletes the stored private key. Open findings from Google Workspace are resolved, and apps only Google Workspace reported leave Discovery. Your reviews of those apps are kept. To cut off access on Google's side too, remove the service account's domain-wide delegation, or delete its key.

> [accordion] Troubleshooting
>
> | Message | What to do |
> | - | - |
> | Administrator address must be an email address | Enter a full email address |
> | Service account address must be an email address | Paste the `client_email` from the JSON key |
> | Google rejected the service account. Check the private key, and that the client id is granted these scopes in Admin console > Security > API controls > Domain-wide delegation. | Usually a missing delegation or scope. Check all three scopes are added for the service account's client id |
> | The private key is not a PEM RSA private key | Paste the whole `private_key` value from the JSON key |
> | Google Workspace refused access: the credentials need admin.reports.audit.readonly (or another scope) | Add the scope named in the message, and check the administrator address belongs to a super administrator |
> | Google Workspace is rate limiting this tenant; try again shortly | Wait, then choose **Check again** |

## Jamf

![Jamf](https://mintcdn.com/xenovia/RO5UF1Tk8L9B7yPt/images/logos/jamf.png?fit=max\&auto=format\&n=RO5UF1Tk8L9B7yPt\&q=85\&s=b5ce2cef727a0513e2ffb773f36651d0)

The Jamf integration reads your Jamf Pro computer inventory. Each AI app installed on a managed Mac shows up in [Discovery](https://docs.xenovia.io/discovery/overview.md), with the Macs it is on and the people who use them, and each Mac gets its own page. Jamf is a detection source: it shows that an app is installed, but can't govern what happens inside it.

When nothing else can govern an app, Xenovia can remove it with Jamf for you. Removing is optional and needs an extra privilege.

### Before you start

In Jamf Pro, create an API role and an API client that uses it. Give the role:

| Privilege | Why |
| - | - |
| Read Computer Inventory | List your Macs and the apps on them |
| Read Smart Computer Groups | Optional: report which Macs you hold compliant |
| Create Restricted Software | Optional: lets Xenovia remove an app |

Then create a client secret for the API client.

### Connect Jamf

**1. Open the integration**

In **Settings › Integrations**, find **Jamf** under **Device management** and choose **Connect**.

**2. Enter your details**

| Field | What to enter |
| - | - |
| **Jamf Pro URL** | Your Jamf Pro address, for example `https://acme.jamfcloud.com` |
| **API client id** | The API client's id |
| **API client secret** | The API client's secret |
| **Compliant smart group** | Optional. The smart group whose members you treat as compliant |

**3. Connect and check**

Choose **Connect**. Xenovia gets a token and checks that it can read your Jamf Pro version and computer inventory, and the smart group if you named one. If a required check fails, Xenovia shows why and saves nothing.

Jamf has no compliance verdict of its own, so the **Compliant smart group** is your answer. A Mac outside that group is reported as not compliant, and an AI app on such Macs gets a finding. Leave it empty and no Mac is called either.

Xenovia keeps the client secret encrypted and never shows it again. It uses the secret only to get short-lived tokens from Jamf Pro.

### What Xenovia reads

Every 6 hours, Xenovia reads each managed Mac's general, hardware, user and application details. It keeps only the installed apps its AI app catalog recognizes, with the bundle id and process name. It only reads; it changes nothing in Jamf unless you ask it to remove an app.

### Remove an app

Open the app in Discovery and choose **Plan a fix**. When nothing else can govern the app, the plan can propose **Remove it with Jamf**. Applying it creates a Restricted Software record in Jamf Pro for the app's process, scoped to all computers. Jamf then quits the app on every managed Mac and shows people "This app is not allowed by your organization." The method's **Delete the app, not only quit it** setting decides whether Jamf also deletes the app.

To stop enforcing it, delete the Restricted Software record in Jamf Pro. A deleted app isn't restored.

Xenovia only uses process names that Jamf itself reported for that entry.

### Disconnect

Choose **Disconnect** on the integration page. Xenovia stops syncing and deletes the stored client secret. Its Macs leave Discovery, open findings from Jamf are resolved, and apps only Jamf reported leave Discovery. Your reviews of those apps are kept. Restricted Software records Xenovia created stay in Jamf Pro until you delete them. To cut off access on Jamf's side too, disable or delete the API client.

> [accordion] Troubleshooting
>
> | Message | What to do |
> | - | - |
> | Jamf Pro URL must be a public https\:// address | Enter your Jamf Pro address with `https://` and no path |
> | Jamf Pro rejected the client id or client secret | Check both, and create a new secret if needed |
> | Jamf Pro refused access: the credentials need Read Computer Inventory | Add the privilege to the API client's role |
> | Jamf Pro refused access: the credentials need Read Smart Computer Groups | Add the privilege, or clear **Compliant smart group** |
> | Jamf Pro is rate limiting this tenant; try again shortly | Wait, then choose **Check again** |
> | Jamf Pro refused: the client needs Create Restricted Software | Removing needs Create Restricted Software. Add it, then apply the fix again |
> | Jamf Pro already has a restriction with that name | A Restricted Software record for this app already exists in Jamf Pro |

## Microsoft Intune

![Microsoft Intune](https://mintcdn.com/xenovia/RO5UF1Tk8L9B7yPt/images/logos/intune.svg?fit=max\&auto=format\&n=RO5UF1Tk8L9B7yPt\&q=85\&s=e64be4a5c878b61404660b759338a4ef)

The Microsoft Intune integration reads your managed devices and the apps Intune detected on them, through Microsoft Graph. Each AI app installed on a managed device shows up in [Discovery](https://docs.xenovia.io/discovery/overview.md), with the devices it is on and the people who use them, and each device gets its own page. Intune is a detection source: it shows that an app is installed, but can't govern what happens inside it.

Xenovia also reads whether Intune holds each device compliant and, for devices running an AI app, which policy setting they are failing.

When nothing else can govern an app that Intune deploys, Xenovia can have Intune uninstall it. Removing is optional and needs an extra permission.

### Before you start

In Entra, register an application and give it these Microsoft Graph **application** permissions, with admin consent:

| Permission | Why |
| - | - |
| `DeviceManagementManagedDevices.Read.All` | List managed devices and the apps detected on them |
| `DeviceManagementConfiguration.Read.All` | Name the compliance policy setting a device is failing |
| `DeviceManagementApps.ReadWrite.All` | Optional: lets Xenovia remove an app Intune deploys |

Then create a client secret for the application. You can use the same app registration as [Microsoft Entra ID](#microsoft-entra-id) if it has all the permissions both need.

### Connect Intune

**1. Open the integration**

In **Settings › Integrations**, find **Microsoft Intune** under **Device management** and choose **Connect**.

**2. Enter your details**

| Field | What to enter |
| - | - |
| **Tenant id** | The directory (tenant) id, or its domain |
| **Application (client) id** | The registered application's client id |
| **Client secret** | The secret you created |

**3. Connect and check**

Choose **Connect**. Xenovia gets a token for the application and checks that it can read managed devices, detected apps and compliance settings. If it can't read devices or apps, Xenovia shows why and saves nothing.

Without `DeviceManagementConfiguration.Read.All` the connection still works. The check names the missing permission, and only the compliance details stop.

Xenovia keeps the client secret encrypted and never shows it again. It uses the secret only to get short-lived Graph tokens.

### What Xenovia reads

| What | How often |
| - | - |
| Managed devices, the AI apps Intune detected on each, and whether each is compliant | Every 6 hours |
| Which policy settings the devices running an AI app are failing | Every 12 hours |

Xenovia keeps only the apps its AI app catalog recognizes and skips the rest. When devices running an AI app fail a compliance setting, the app gets a finding that names the setting. Xenovia only reads; it changes nothing in Intune unless you ask it to remove an app.

### Remove an app

Open the app in Discovery and choose **Plan a fix**. When nothing else can govern the app, the plan can propose **Remove it with Intune**. Applying it finds the app's record in Intune by name and adds an **uninstall** assignment for all devices. Xenovia keeps the app's existing assignments, so a deliberate deployment to a group isn't undone. Where the assignments disagree about a device, Intune's own precedence decides.

Intune can only uninstall an app it deploys. An app that Intune only detects has no such record, and Xenovia refuses to remove it.

To undo it, remove the uninstall assignment from the app in Intune.

### Disconnect

Choose **Disconnect** on the integration page. Xenovia stops syncing and deletes the stored client secret. Its devices leave Discovery, open findings from Intune are resolved, and apps only Intune reported leave Discovery. Your reviews of those apps are kept. Uninstall assignments Xenovia added stay in Intune until you remove them. To cut off access on Microsoft's side too, delete the client secret or the app registration.

> [accordion] Troubleshooting
>
> | Message | What to do |
> | - | - |
> | Tenant id must be the directory id or its domain | Enter the tenant id, or a domain such as `acme.onmicrosoft.com` |
> | Microsoft rejected the tenant id, client id or client secret | Check all three, and that the client secret hasn't expired |
> | Microsoft Intune refused access: the credentials need DeviceManagementManagedDevices.Read.All | Add the permission as an application permission, and grant admin consent |
> | Microsoft Intune is rate limiting this tenant; try again shortly | Wait, then choose **Check again** |
> | Intune does not deploy an app called *name*: it only detects it, and a detected app cannot be uninstalled from here | Intune can't uninstall an app it doesn't deploy. Plan a different fix |
> | Intune refused: the app needs DeviceManagementApps.Read.All (or ReadWrite.All) | Removing needs `DeviceManagementApps.ReadWrite.All`. Add it with admin consent, then apply the fix again |
> | Intune has no app with that id | The app record was removed from Intune. Sync again |

## What next?

> [card] [Discovery](https://docs.xenovia.io/discovery/overview.md)
>
> Review what the integrations found.

> [card] [GitHub](https://docs.xenovia.io/discovery/github.md)
>
> Find the agents in your code too.

> [card] [AI platforms](https://docs.xenovia.io/agents/ai-platforms.md)
>
> Govern Claude and OpenAI platforms your teams use.


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