# Employee Setup

Employee Setup runs the first time you sign in. The flow has two steps. Plan on under 5 minutes end-to-end.

***

### 📖 **What you'll need**

* The invite email from your admin.
* Sign-in credentials for the apps you'll connect (Slack, GitHub, Jira, Linear and so on).
* Your Agent client installed locally (Cursor, Claude, Copilot, ChatGPT or another supported client).

:bulb: *Why this matters:* Cakewalk routes every tool call your Agent makes through the MCP Gateway. Setup wires your Connections to your identity and your Agent to the gateway.

***

### 1️⃣ Step 1: Set Up Connections

Connect to the apps your Agent will access.

1. Open **My Connections**. You'll see a card for every Connection your admin has made available.
2. Find the Connection you need and click **Connect** on the card.
3. Cakewalk redirects the full page to the app's sign-in screen. Sign in and authorize.
4. You return to **My Connections**. The card flips to **Connected**, the connected date appears and the **Connect** button is replaced by a 3-dot menu.

{% hint style="info" %}
You can skip Step 1 and connect later from the same grid. Your Agent only accesses Connections you've authorized.
{% endhint %}

**Card statuses you may see**

* **Connected**: ready to use.
* **Not Connected**: admin made it available, you haven't connected yet. Default starting state.
* **Error**: auth or transport issue. Sign in again to reconnect.

:bulb: *Why this matters:* Connections are per-User. Your Agent inherits exactly your access in each app.

***

### 2️⃣ Step 2: Set Up Your Agent

Wire your Agent to Cakewalk's MCP Gateway.

1. Open **My Agents**. You'll see a card for every Agent in the company catalog.
2. Find your Agent and click **Set up**.
3. The setup modal renders the install method for your specific Agent. Follow the one shown:
   * **Deep link**: click **Open in \[Agent name]**. Your Agent client launches with the configuration pre-filled.
   * **JSON config**: copy the JSON block and paste it into your Agent's config file. Per-OS toggle (macOS, Windows, Linux) shows the right path. Confirm with **I've added it**.
   * **Custom form**: follow the named, screenshot-illustrated steps for your Agent (for example, Claude Desktop's MCP setup screen). Copy the values shown.
   * **CLI**: copy the command, run it in your terminal and click **I've run it**.
4. Once your Agent connects, the card flips to **Active** and shows session count and last active.

{% hint style="info" %}
If your admin adds an Agent for you with **Add a seat**, the Agent appears as **Inactive** in your **My Agents** until you run setup.
{% endhint %}

**Card statuses you may see**

* **Active**: your Agent is wired to Cakewalk and ready.
* **Inactive**: Agent is in your catalog, not set up yet. Default starting state.
* **Not connected**: your Agent previously connected and was disconnected.

:bulb: *Why this matters:* One setup per Agent. After that, every tool call your Agent makes routes through Cakewalk and gets evaluated against your company's Policies.

***

### ✅ What Happens After Setup

* Your Agent's tool calls go through the MCP Gateway, with credentials mediated server-side. Your Agent never sees your tokens.
* Default Policies act per action type. Approvals surface as inline prompts in your Agent client (via MCP elicitation):
  * **Read** = auto-approves except on Confidential Connections (approval required).
  * **Write** = approval required.
  * **External** = approval required.
  * **Destructive** = denied.
* You can see what your Agents did in **My Connections** and **My Agents** at any time.

***

### 👉 **Where to go next**

* Open your Agent and run a normal task. Watch the inline approval prompts when they appear.
* Need to disconnect a Connection? Open the 3-dot menu on the card and select **Disconnect**.
* See the bigger picture: [Cakewalk for Employees](/docs/ai-agent-access/introduction/cakewalk-for-employees.md).


---

# Agent Instructions: Querying This Documentation

If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.

Perform an HTTP GET request on the current page URL with the `ask` query parameter:

```
GET https://www.cakewalk.security/docs/ai-agent-access/introduction/get-going-with-agent-access/employee-setup.md?ask=<question>
```

The question should be specific, self-contained, and written in natural language.
The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.