Understanding the permissions an app requests Guide

When you connect one of your tools to Praxivara, the provider — Google, Microsoft, Slack, or another app — shows you a consent screen before the connection completes. This page explains what that screen is asking for, why it appears, and how you stay in control of exactly what your assistant and agents can do.

Why a consent screen appears

Praxivara connects to your own account for an app so the assistant can work inside it on your behalf. Before that link is created, the app itself asks you to confirm the connection. This screen is presented by the provider, not by Praxivara, which is why its wording and appearance vary from one app to the next.

The permissions listed describe the kinds of information and actions the connection could involve — for example, reading and sending email, viewing your calendar, or accessing your contacts. Granting them establishes the link; it does not, on its own, cause anything to happen in your account.

Note: A connection is private to you and scoped to your workspace. Other members of your organization do not gain access to an app because you connected it.

What the permissions typically cover

The exact list depends on the app, but requests generally fall into a few familiar groups:

• Reading data — viewing messages, events, records, files, or contacts so the assistant can find and summarize information.
• Creating or updating data — drafting or sending email, adding calendar events, updating CRM records, or creating documents.
• Account identity — confirming which account you are connecting, so the right inbox, calendar, or workspace is linked.

Broad-sounding permissions are common because providers describe access in wide categories. An app may ask for the ability to manage email even if you only intend the assistant to read it. That is expected — the next step is where you narrow things down.

Permissions on the provider screen vs. actions in Praxivara

There are two layers of control, and this distinction matters:

1. The provider consent screen establishes what the connection is technically permitted to reach.
2. The action list in Praxivara determines what your assistant is actually allowed to do with that connection.

When you connect an app, Praxivara shows you its available actions as a checklist. You choose exactly which ones to enable — you can select everything, or turn on only the specific actions you trust. This lets you connect a powerful app while limiting the assistant to, for example, reading only.

How to review and limit what the assistant can do

1. On the Integrations page, open the app you want to connect and select Connect.
2. Complete the provider's sign-in or authorize screen, reviewing the permissions it lists.
3. Back in Praxivara, review the list of available actions for that connection.
4. Enable only the actions you want the assistant to perform, or choose the option to select all.
5. Finish connecting and confirm the app now shows as connected on the hub.

You can revisit these choices at any time. Open the connection's Manage actions view to add or remove enabled actions as your needs change. If the assistant ever reports that a specific action is not enabled, this is where you turn it on.

Tip: To keep a connection read-only, enable only the actions that view or retrieve information and leave the ones that create, send, or update turned off.

If you are unsure about a permission

You are never committed by opening the consent screen — you can close it and the connection simply will not be created. If a requested permission looks broader than you expected, remember that the action checklist in Praxivara is your safeguard: even a wide-reaching connection only ever performs the specific actions you have enabled.

If you would rather not grant a certain kind of access at all, you can connect the app with a narrower set of actions, or choose not to connect it. You remain in control at every step.