# Admin

> What owners and admins manage, and where each setting lives.

**Admin** in the sidebar is for the owner and admins. It opens on **Home**, which lists what needs attention first.

![Admin Home: Needs attention at the top, then one card per area, with the Admin menu on the left](../images/admin/home.webp "Admin Home")

## The Admin menu

| Group | Pages |
|---|---|
| **People** | Members & invites, Groups, Folder permissions |
| **Shared with the team** | AI engines, AI connections, MCP, Plugins |
| **Activity** | Usage, Activity log |
| **Settings** | Team settings, System |

## Your first week

1. **Set permissions**

   Assign folders to groups and check access with two test accounts. [Read about permissions](/docs/use/files-and-permissions.md).

2. **Connect the AI**

   The owner connects an authorized Claude or OpenAI account in **AI connections**. **AI engines › Providers** shows whether each engine is ready.

   ![Admin › AI engines › Providers: Claude Code, Codex and the API engine, each with its state, models, check again and an on/off switch](../images/admin/engines.webp "AI engines › Providers")

3. **Invite the team**

   Start with a small pilot group, then confirm roles and folder access.

4. **Add a first skill**

   Copy a starter skill and review its result against the sources. [Read about skills](/docs/use/team-skills.md).

5. **Make a first backup**

   Back up, then practise a restore with IT. [Read about backups](/docs/operate/backups.md).

## What admins cannot do

- Read other people's chats. The activity log shows which files the AI read, not the conversation.
- See personal folders or personal skills in `users/<id>/`.
- Use a team MCP connection, unless a rule names them.

> **Warning:** People who operate the server, its OS or its storage can read stored files and chats. In hosted plans, that includes Trina's operators.

## AI engines

**Admin › AI engines** has two tabs.

| Tab | What it holds |
|---|---|
| **Providers** | One row per engine (Claude Code, Codex, API): its state, sign-in mode, models and last check, with **↻** to check again and an on/off switch. |
| **Team provider settings** | The default engine and model, answers at the same time, who may pick the engine, switching to Codex at Claude's limit, and usage limits. |

Click an engine's name to open its details: what is missing and how to fix it, every model, and for the owner the API engine form.

![An engine row opened: the reason it is not ready, its sign-in mode, models and last check](../images/admin/engines-open.webp "An engine opened")

![Admin › AI engines › Team provider settings: team defaults, who picks the engine, the switch to Codex, and usage limits](../images/admin/engines-team.webp "Team provider settings")

Save and Cancel appear in a section only after you change something in it.

> **Note:** The two tabs are new in 0.4.0. In 0.3.0 and earlier, everything is on one **AI engines** page.

## Activity log

**Admin › Activity log** lists sign-ins, admin actions, file changes and AI tool use, grouped by day. The filters stay at the top while you scroll. It never shows chat content.

![Admin › Activity log: filters for kind, person and time, then one line per event, grouped by day](../images/admin/activity-log.webp "The activity log")

## System

**Admin › System** has backups, diagnostics and updates. Diagnostics shows problems such as a missing engine, a failed check or disabled automatic updates.