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

# Getting started checklist

> See which steps are left after Setup, which apply to your product, which are done and which need attention.

After [Setup](https://agentrail.co/docs/guides/setup.md), the **Getting started** page lists what is left to install or connect: what records agents on your product, and what lets Agentrail do more with it. Each step is worked out from what your workspace already has, every time the page opens, so a step moves to **Done** on its own once its first event or record arrives.

Open it from the **Getting started** section at the bottom of the sidebar, above your account. Its header shows how many of the steps that apply are done, and how many need attention. Expand it to see each step with its state, and select one to open it on the page. The section is hidden once every step that applies is done, or once someone closes it. The page stays in page search: select **Search** in the sidebar, or press Command K (Control K on Windows and Linux), and type **Getting started**.

When the sidebar is collapsed, the section is a ring that fills as steps are done. Hover it to see the count, and select it to expand the sidebar.

## Read the page

The steps that apply are listed on the left, under how many of them are done. Select a step to read, on the right, what it gives your workspace, where it stands, and the button that opens the page where it is done. The address keeps the selected step, so you can send someone a link to it.

Each step that applies has one of these states:

| State | Meaning |
| - | - |
| **Done** | What the step asks for exists. |
| **Waiting** | It is installed or started, and its first evidence has not come. |
| **Needs attention** | It is set up, and something says it fell short. |
| **To do** | Nothing has started. |

Below the list:

* **Waiting on research** lists steps your Brand Vault decides, until company research publishes it.
* **Not needed** lists steps that do not apply, each with the reason. When the reason is something your Brand Vault lists, or does not, an Admin sees **Correct it in the Brand Vault**, to fix what research got wrong.

## The steps

| Step | Applies when | Done when |
| - | - | - |
| **Research your company** | Always | The Brand Vault has its first version. It needs attention if research failed. |
| **Connect GitHub** | Always (recommended) | GitHub is connected. It needs attention if GitHub paused the connection. |
| **Record every request** | Always | An edge, Cloudflare logs or Vercel logs source has sent an event. It needs attention while only the browser script has sent events. |
| **Add the browser script** | Your Brand Vault lists a website or web app, or you added a browser script | A browser script has sent an event. |
| **Add your servers** | Your Brand Vault lists an API, web app, sign-up or MCP server, or you added a server source | A server source has sent an event. |
| **Record sign-in** | Your Brand Vault lists a sign-up or web app, or a source that has sent events reports sign-in | A source that has sent events reports sign-in. |
| **Record MCP calls** | Your Brand Vault lists an MCP server, or a source that has sent events reports MCP calls | A source that has sent events reports MCP calls. |
| **Run a readiness scan** | Agent readiness is on for the workspace | The workspace has run a scan. |
| **Set up agent feedback** | Always | A feedback project that is accepting reports has a placement or has received a real report. A test report doesn't count here. |
| **Send a test report** | Always | A feedback project has received a report, a test or a real one. |
| **Connect your own agent** | Your organization allows agent access | You have connected an agent. It counts from its first call. |

What you have installed always counts: a source you added makes its step apply, whatever the Brand Vault says and before research finishes. For **Record sign-in** and **Record MCP calls**, that is a source that has sent events and reports what it captures. A source counts only until it is revoked.

**Connect your own agent** is the only personal step. Every other step reads the same for everyone in the workspace.

## Close the section

Select the close button in the section's header. It closes the section for everyone in the workspace, and anyone in the workspace can close it, Members included. A step that needs attention later does not bring it back. The **Getting started** page stays in page search.

## Admins and Members

Everyone in the workspace sees every step. A Member can run a readiness scan, send a test report and connect their own agent. Every other step is set up by an Admin: a Member sees its state and **An Admin sets this up.** in place of its button.


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