What is Assrt (asrt)?
People type asrt for three different things. If you landed here from a developer corner of the internet, you almost certainly meant Assrt, the testing tool. The doubled s is easy to drop, so the name gets shortened a lot. Here is the straight answer, then the part no other page on the abbreviation will tell you: the exact syntax you write and what runs when you do.
Assrt (spelled a-s-s-r-t, at assrt.ai) is a free, MIT-licensed AI testing framework. You describe test cases in plain English; Assrt generates and runs them as real Playwright tests in a real browser, then reports pass or fail per case. It installs as a CLI and as an MCP server for coding agents.
Two unrelated things also go by ASRT: the American Society of Radiologic Technologists (a medical-imaging association) and the Nasdaq ticker ASRT for Assertio Holdings. This page is about the testing tool.
Three meanings, one spelling
Worth clearing up first, because the search term collides. Only one of these is a piece of software you install.
Assrt (assrt.ai)
Open-source AI testing framework. Turns plain-English #Case blocks into real Playwright runs. MIT licensed, free, runs on your own CI. This is the one developers mean.
ASRT (asrt.org)
American Society of Radiologic Technologists. A professional association for medical imaging and radiation therapy, based in Albuquerque. Nothing to do with software.
ASRT (Nasdaq)
Stock ticker for Assertio Holdings, a pharmaceutical company. A finance symbol, not a product.
The whole tool, in one block of text
Most explanations of an AI test tool stop at the pitch. Here is the actual thing you write. Assrt reads a plain-English format it calls a #Case block. Each #Case is one scenario. You give it a title and a sentence or two of intent. Assrt parses each block, drives a real browser to execute it, and reports pass or fail per case.
assrt run --url http://localhost:3000 --plan "
#Case 1: Homepage loads
Navigate to the homepage and verify the heading is visible.
#Case 2: Login works
Click Sign In, enter test@example.com / password123,
verify the dashboard appears.
"No selector, no assertion library, no page-object file. The two #Case blocks above are the entire test. Assrt figures out the selectors and the waits, runs both scenarios in a real browser, and prints a result for each. Swap --plan for --plan-file tests.txt to keep them in a file, or add --json for CI.
What happens after you hit enter
The loop Assrt runs for every #Case, from a blank file to a green check.
The Assrt run loop
Discover
The three tools an agent gets
Assrt installs as a Model Context Protocol server, so a coding agent like Claude Code or Cursor can call it directly. Setup exposes exactly three tools. The most-used one, assrt_test, runs a scenario and reports back. Here is the round trip.
An agent runs assrt_test
Runs #Case scenarios against a URL and returns structured pass or fail.
Navigates a URL and auto-generates test cases so you do not start cold.
Reads a failed run, finds the root cause, and suggests a corrected test.
Where it plugs in
Because Assrt is an MCP server plus a plain CLI, it sits next to the editor or agent you already use rather than asking you to move into a new dashboard.
What makes it worth a name of its own
There are hosted services that also generate browser tests. The difference that earned Assrt is ownership. The output is standard Playwright living in your repo, the source is MIT licensed, and the whole thing runs on your CI with no cloud dependency you cannot remove. Nothing is locked behind a proprietary format you would have to re-author on the way out.
Free to start, free to stay. MIT licensed. Tests live in your repo.
Closed source, tests stored inside the vendor, billed per seat or per run.
Want to see Assrt run against your app?
Bring a real flow and we will turn it into a passing #Case in a live browser on the call.
asrt, answered
Is asrt the same thing as Assrt?
On this site, yes. Assrt is spelled a-s-s-r-t and is frequently typed as 'asrt' because the doubled 's' is easy to drop. Assrt is a free, open-source AI testing framework that generates real Playwright tests. The abbreviation ASRT in all caps usually refers to the American Society of Radiologic Technologists or the Nasdaq ticker for Assertio Holdings, which are unrelated to this tool.
What does Assrt actually generate?
Standard Playwright. You describe test cases in plain English using a #Case block, and Assrt drives a real browser to execute them, returning structured pass or fail results with screenshots and an optional video recording. Because the output is ordinary Playwright running in a real browser engine, there is no proprietary file format to learn and nothing to export later.
What is the #Case format?
It is the plain-English test syntax Assrt reads. You write '#Case 1: Homepage loads' followed by a sentence or two describing the steps and what to verify, then '#Case 2:' for the next one. Assrt parses each #Case block, runs it as its own scenario, and reports pass or fail per case. You can pass the blocks inline with --plan or from a file with --plan-file.
Do I need to know Playwright to use Assrt?
No. The whole point of the #Case format is that you write what a test should check in English, not in code. Assrt handles the selectors, waits, and assertions. If you do know Playwright, the generated runs are still standard, so you can read and reason about exactly what executed.
Is Assrt free and open source?
Yes. Assrt is MIT licensed and free to start and free to stay. The source for the SDK and MCP server is public on GitHub. Comparable hosted QA automation services run into the thousands of dollars per month and keep your tests inside their platform; Assrt keeps the tests in your repo and runs them on your own CI.
How does Assrt work inside Claude Code or Cursor?
Assrt ships an MCP server that exposes three tools to coding agents: assrt_test runs scenarios against a URL, assrt_plan navigates a URL and auto-generates test cases, and assrt_diagnose analyzes a failed run and suggests a corrected test. After setup, an agent can write a feature and then verify it in a real browser without you leaving the editor.
Can Assrt reuse my existing logged-in browser session?
Yes. Extension mode connects to your running Chrome with its existing cookies and logins instead of launching a fresh browser, so you can test flows behind authentication. The first run walks you through a token, then auto-saves it to ~/.assrt/extension-token so later runs are seamless.
Keep reading
AI-generated Playwright E2E tests
How a model turns a described flow into a Playwright run, and where the generated code holds up versus where it needs a human pass.
Playwright vs Selenium
The browser-engine and waiting-model differences that decide which framework your suite should sit on.
Self-healing vs static selectors
What 'self-healing' really does when the DOM moves, and when a static locator is still the safer choice.
Comments (••)
Leave a comment to see what others are saying.Public and anonymous. No signup.