> ## Documentation Index > Fetch the complete documentation index at: https://support.i.moneyforward.com/llms.txt > Use this file to discover all available pages before exploring further. # Permission Activity overview > Learn how to use the Permission Activity screen in AI Helpdesk to review the history of tool approvals and executions by the AI agent, and of permission and integration setting changes. The **Permissions** > **Permission Activity** screen shows the history of events for the tools that the AI agent uses. These events cover tool permissions and tool executions. It records approvals and denials of tool executions and the tools that the AI agent runs. It also records changes that admins make to permission and integration settings. ## Open the screen In the left navigation, open **Permissions** > **Permission Activity**. * Only logged-in users with the **Admin** role can view the Permission Activity screen * Logged-in users without the **Admin** role do not see **Permissions** in the left navigation. If they open the screen's URL directly, they are taken to the **Home** screen ## List columns The list shows events with the newest first. Click a column header to sort by that column. | Column | Description | | - | - | | **Timestamp** | The date and time the event was recorded | | **Category** | **Human Operations** or **Guardrail Decisions** (see [Recorded events](#events)) | | **Event Type** | One of **Permission configured**, **Integration configured**, **Permission Approval**, or **Execute Approval** | | **Actor** | The person who caused the event. Besides the name of a logged-in user, it can show **AI HelpDesk**, **System**, or **Requester** | | **Tool** | The name of the tool. For integration events, it shows the name of the MCP server (for example, **Admina MCP**) | | **Tier** | The type of tool. One of **Read**, **Update**, **Delete**, or **Unclassified** | | **AI Integration** | The name of the MCP server that provides the tool (for example, **Admina MCP**) | | **Outcome** | The result of the event (see [Outcomes](#outcomes)) | In **Permissions** > **Policies**, **Read** matches **Read-Only Tools** and **Update** matches **Update-Related Tools**. **Delete** matches **Delete Tools**. For integration events, the **Tier** does not show a type of tool. A warning icon next to the **Outcome** means that the requester decided the request themselves (see [Self-approval](#self-approval)).

Recorded events

Events are grouped by **Category** and **Event Type** as follows. | Category | Event Type | When it is recorded | | - | - | - | | **Guardrail Decisions** | **Permission Approval** | When the AI agent asks for approval to run a tool whose permission is **Ask**. It is also recorded when the request is approved or denied, when it expires, and when closing the session cancels it | | **Guardrail Decisions** | **Execute Approval** | When the AI agent runs a tool. For tools set to **Allow**, it is recorded each time the tool runs. For tools set to **Ask**, it is recorded when the request is approved and the AI agent resumes | | **Human Operations** | **Permission configured** | When you click **Save Changes** in **Permissions** > **Policies** | | **Human Operations** | **Integration configured** | When you save credentials or turn **Enabled** on or off in the settings dialog of an MCP server on the **Settings** > **Integration** tab | The AI agent does not use tools set to **Deny**, so no events are recorded for them. To set tool permissions, see [Admina MCP server integration](/en/ai-helpdesk/integrations/admina#tool-permissions). **Permission configured** events are recorded one per tool each time you save. They cover tools set to **Ask** and tools whose permission differs from the default. As a result, tools that you did not change in that save can also appear.

Outcomes

| Outcome | Event Type | Meaning | | - | - | - | | **Pending** | **Permission Approval** | The AI agent asked for approval, and nobody has approved or denied it yet | | **Approved** | **Permission Approval** | An approver approved the execution | | **Denied** | **Permission Approval** | An approver denied the execution | | **Expired** | **Permission Approval** | Nobody approved or denied the request within 30 minutes | | **Cancelled** | **Permission Approval** | The session was closed while the request was waiting for approval | | **Executed** | **Execute Approval** | The AI agent ran the tool | | **Configured** | **Permission configured** | Tool permission settings were saved | | **Connected** | **Integration configured** | The settings were saved while the MCP server was enabled | | **Removed** | **Integration configured** | The settings were saved while the MCP server was disabled | One approval flow records more than one event. For example, an approved request records three events: **Pending**, **Approved**, and **Executed**. The **Actor** depends on the event. | Event | Actor | | - | - | | Asking for approval (**Pending**) | **Requester** | | Approving or denying (**Approved**, **Denied**) | The logged-in user who approved or denied | | Expiring (**Expired**) | **System** | | Cancelling because the session was closed (**Cancelled**) | The person who closed the session. **System** if the session was closed automatically | | Running a tool (**Executed**) | **AI HelpDesk** | | Saving permission or integration settings (**Configured**, **Connected**, **Removed**) | The logged-in user who saved | Approvers approve or deny requests in places such as the ticket detail panel. For how tickets that wait for approval appear, see [Ticket list and details](/en/ai-helpdesk/operations/tickets#status-column). ## Filter events Use the filter bar above the list to narrow down the events. | Filter | What it matches | | - | - | | Search box | Events whose **Actor** or tool name contains the text you enter | | **All categories** | **Human Operations** or **Guardrail Decisions** | | **All event types** | **Permission configured**, **Integration configured**, **Permission Approval**, or **Execute Approval** | | **All tiers** | **Read**, **Update**, **Delete**, or **Unclassified** | | Period | **Last 7 days**, **Last 30 days**, **Last 90 days**, or **All time** (the default is **All time**) | To remove the filters, click **Clear all filters**. **Clear all filters** appears while any filter is in use. When no events match, it also appears in the list. ## View event details Click a row in the list to open the event details on the right side of the screen. The top of the details shows the tool name and the **Category**, **Tier**, and **Outcome**. **Entry Details** shows the following items. | Item | Description | | - | - | | **Timestamp** | The date and time the event was recorded | | **Actor** | The person who caused the event | | **Event Type** | The type of event | | **AI Integration** | The MCP server that provides the tool | | **Session ID** | The ID of the session the approval belongs to (for Guardrail Decisions events, when recorded) | | **Target Resource** | The target of the tool's operation (for Guardrail Decisions events, when recorded) | Some events also show the following sections. | Section | Events | Description | | - | - | - | | **Current Value** | Human Operations | The saved setting. For **Permission configured**, the permission and approval routing. For **Integration configured**, whether the server is enabled | | **Parameters** | Guardrail Decisions | The values the AI agent tried to pass to the tool | | **AI Reasoning** | Guardrail Decisions | Why the AI agent tried to use the tool | **AI Reasoning** appears on events for asking for approval and for approving or denying. **Parameters** also appears on events where the AI agent ran an approved tool. To close the details, click the close button at the top left, or click **Permission Activity** in the breadcrumb.

Self-approval

Some tools use **Requester** as the approval routing. When the requester approves or denies a request for such a tool, the details show **Self-approved** at the top. In the list, a warning icon appears next to the **Outcome**. You can choose **Requester** only when **Self-Approval** is enabled on the **Settings** > **Admin** tab (see [Choose who can approve](/en/ai-helpdesk/integrations/admina#approval-routing)). ## Related pages * [Approve tool runs](/en/ai-helpdesk/permissions/approvals) * [Admina MCP server integration](/en/ai-helpdesk/integrations/admina) * [Ticket list and details](/en/ai-helpdesk/operations/tickets) This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.