Getting started

This page covers setting up your account, signing in, finding your way around and running your first test.

Set up your account

Accounts are created by invitation: an Admin of your organization adds you and sends you a setup link (see Members). Self sign-up and email notifications are not available yet.

Open the link, choose a display name and a password, and click Create Account. The password needs at least 8 characters, with an uppercase letter, a lowercase letter and a number. A setup link is valid for 7 days; if it has expired, ask your Admin to regenerate it.

Sign in

Sign in at cgt.codemodity.com  with your email and password. Reelspect opens the Runs page of the last project you worked in.

Find your way around

• The sidebar shows one scope at a time. Project pages are Library, Runs, Sandbox, Knowledge and Settings. Organization pages (reached through Projects or Manage organization in the organization switcher) are Settings (members and integrations) and Devices.
• Bridges you registered are under My Bridges in the user menu (your avatar). See Device Bridge.

Resource Monitor

The ring on the right of the header shows how busy the test agents are. Click it to list the running and queued work in the organization. Sessions wait in a queue while capacity is busy and start automatically.

• Hover an item and click Cancel to stop it. Cancelling a run’s work needs the Tester or Manager role in its project.
• Organization Admins also see Stop all: it cancels every running and queued run and Sandbox session in the organization. Schedules are not paused.

Your first test

A test case moves through three places, each with its own purpose:

• Sandbox: get the test right. A new prompt rarely works on the first try: the agent may misread a step, stop too early or judge the result differently from how you would. The Sandbox is private, so you can launch, read the result, rewrite and launch again as often as you need without anyone else seeing half-finished tests or failed results.
• Library: share what works. Once a test case behaves reliably, promote it. The Library is the project’s shared, curated set of test cases that everyone builds runs from. Keeping drafts out of it means anyone picking test cases for a run can trust that they work.
• Runs: test the product. A run executes a chosen set of Library test cases against one target, with shared settings (device, URL, instructions), and keeps the results. Runs can be repeated for every release or put on a schedule, so their results are the ones that tell you whether the product works, not whether the prompt does.

This walkthrough takes one test through all three. Promoting and creating runs need the Tester role or higher in the project; see Roles.

1. Draft it in the Sandbox. Open Sandbox and click + Test Case. Write the Test Instructions (Prompt): what the agent should do and what counts as a pass (see Writing test cases). For a first test, pick a size from the Desktop group as the Device Preset, and set Source to URL with the casino page or game launch URL.
2. Launch and watch. Click Launch (or the play icon on the row). While it runs, the eye icon (View live session) opens the Session viewer. When it ends, open Execution History from the row menu to read the verdict and the agent’s explanation. If the result is not what you expected, sharpen the instructions and launch again.
3. Promote it to the Library. Choose Promote to Library from the row menu, then Create New Test Case and a Target Collection. Your Sandbox copy stays in the Sandbox. The Sandbox Source URL is not copied, so the run supplies it in the next step.
4. Add it to a run. On Runs, create a plan with + Plan, select it and click + Run. Choose a Device Preset and a Source URL, and under Test Cases choose Select specific cases and pick your test case. Select the run and click Launch. When it finishes, click the test in the Tests list to see its verdict, then View Session to replay it.

Next, read Runs for execution strategies and rounds, and Schedules for automatic launches.