Skip to main content

What is the TestSprite Web Portal?

The TestSprite Web Portal is where you work with TestSprite’s agent: a QA engineer that joins your project, explores your app, writes and maintains its test cases, runs them, and tells you what broke and why. You talk to it in plain language and approve what it proposes; you never write a line of test code. It complements the TestSprite CLI and the MCP Server, which bring TestSprite into your terminal, IDE, and coding agent, by giving your whole team a shared place to set up projects, run tests, and review results.
A project open in the portal: the chat with the agent on the left, the Plan tab beside it with 16 proposed tests
TestSprite is spec-driven: tests come from what your product is supposed to do (your PRD, specs, tickets) and what the live app actually does (what the agent sees when it explores). A PRD isn’t required, but it is strongly recommended. It’s the biggest lever on test quality.

What It Tests

How It Works

1

Tell the agent what to test

Click New Project and paste your app’s address. Attach a PRD (strongly recommended), add your repository, and give it a test account if your app needs sign-in.
New Project: What should we test?, with the chat box to paste your app's address in
2

Create the project

The agent asks for what only you know, then proposes a project on the Setup tab. Nothing is created until you click Create Project. See Setup.
1 The Ready to create card above the chat box; 2 the Setup tab beside it, with the proposed project
3

Watch it explore

For a web app, the agent walks your live app in a browser you can watch on the Exploration tab. For an API, it makes real calls. What it learns goes into the project’s notes. See Exploration & Notes and API Calls.
The agent exploring: its browser steps in the chat, and the Exploration tab with the live browser and Exploration Notes
4

Review the proposed test cases

The plan arrives as a proposal on the Plan tab. Keep, leave out, or edit each test, then click Apply and Run. See Proposed Test Cases.
The Plan tab with the proposal, grouped by use case, and Apply and Run
5

Run and review

Tests run in the cloud. Each ends Passed, Failed, or Blocked, and the agent judges each failure: your app is wrong, the test is wrong, or something got in the way.
A run in progress
6

Keep it running

Run your tests automatically to catch regressions: on every pull request or push with the GitHub integration, or on a cadence with Test Schedules.
GitHub checks on a push: TestSprite passed alongside CI
For the click-by-click version, start with the quickstart for your project type:

UI Quickstart

Your first frontend project, end to end (~10 min)

API Quickstart

Your first backend project, end to end (~10 min)

Working with the Agent

Sessions. Each conversation with the agent is a session, and every session belongs to one project. A project’s sidebar lists its sessions; click New Session to start another. Sessions are yours: other members don’t see them, but they do see everything the agent saved to the project. Every session starts from the project’s shared notes, so a new one doesn’t re-learn your app.
A project's new session: its sessions in the sidebar, New Session, and the chat box with the model picker
It keeps working when you leave. Closing the tab changes nothing. The session is where you left it when you come back. It asks before it changes anything. The agent reads, explores and drafts on its own. Anything that changes your project or spends credits waits on a card above the chat box:
The Ready to run 16 tests card above the chat box, with Not Now and Run
One card waits at a time. Decline one, and the agent asks what to change. Shortcuts. Type / in the chat box for commands:
The chat box with / typed and its command menu open: /setup, /plan, /run, /failures, /new
Many pages also have buttons that ask the agent for you, such as Find the Cause and Find a Fix on a failed result. Model and credits. Pick the session’s model in the chat box before your first message; some models need a paid plan. The line under the chat box shows the tokens and credits the session has used. A session stops when the workspace runs out of credits, and carries on once you top up. Who can use it. Asking the agent needs edit access to the project. Viewers can browse projects and results.

Who It’s For

  • Developers: skip the test-writing boilerplate and see your changes exercised against real flows in minutes.
  • QA teams: replace handwritten regression suites with a plan the agent derives from your live app, kept running with Test Lists and Test Schedules.
  • Engineering leads: get reproducible reports and severity-ranked failures without building a test-writing team.

Advanced Testing Capabilities

Beyond the core flow above:
  • Flexible sign-in: a test account, a one-time-code login with a TestSprite-managed account, or a saved sign-in for SSO such as Google
  • Auto-Heal: tests recover when the page changes instead of failing
  • Step-by-Step Replay: per-step screenshots and a video of every run
  • Map View: every recording from a run, grouped by status
  • Reusable test code: generated Playwright + Python, exportable to GitHub

What’s Inside the Dashboard

The workspace Home, with Home, All Projects, Test Lists, Test Schedules, Settings and Recent Projects in the sidebar
Inside a project, the chat stays on the left and the project’s pages (Sources, Test Cases, Test Runs, Insights, Settings) open as tabs beside it. See Test Detail Page.

Where to Go Next

Key Terms

The core vocabulary

Testing Lifecycle

How UI and API projects differ, phase by phase

UI Quickstart

Your first frontend project

Join the Community

Discord for users and the team