# Momentic

> An AI QA platform whose agent, Mo, bug-bashes web and mobile apps and keeps the findings as regression tests

- Category: AI Coding / AI SDLC Automation
- Repository: https://github.com/momentic-ai/qa
- Website: https://momentic.ai
- Canonical page: https://ai-tldr.dev/tools/momentic/
- Updated: 2026-11-02

## In short

Momentic is an AI quality-assurance platform from a Y Combinator W24 company. It sits in the AI SDLC Automation category of the AI Coding tools. To get started: install the CLI.

## Overview

Momentic is an AI quality-assurance platform from a Y Combinator W24 company. Its agent, Mo, takes a URL or a flow to check, explores the app in a real browser or a hosted mobile emulator, and reports each bug it reproduces with steps and a recording. Mo works across web, iOS and Android apps; Momentic says desktop support is planned.

Findings do not have to stay one-off. Mo can turn its runs into regression tests, written as plain-English steps and assertions in YAML files that live in your repository. The Momentic CLI runs them locally or in CI, and its exit code can act as a required merge or deploy check. When the UI changes, Momentic relocates controls during runs and proposes test-file repairs for you to review, and it sorts failures into application bugs, flakiness or environment issues with screenshots and traces.

Under the hood, Mo is a custom agent harness. It builds a graph of the app as it explores, forks large product areas into parallel sub-agents, and uses specialised agents for planning, finding elements and reading images and video. A tunnel lets the hosted agents reach private sites and local apps, and Mo can test the preview deployment of a pull request and post its findings as a review. The `qa` CLI installs from npm and needs a Momentic account; the platform is free to start, new signups get $250 in Mo spend, and paid and enterprise plans add capacity.

## What it does

- Mo explores web, iOS and Android apps on its own and files bugs with reproduction steps and a recording
- Pull-request testing against preview deployments (Vercel and Cloudflare), with findings posted as a review
- Plain-English tests in YAML files in your repo, run by the CLI locally or in CI as a merge or deploy gate
- Auto-healing locators plus failure triage into app bugs, flakiness or environment problems
- Start sessions from the `qa` CLI, the web app, a PR comment, or a coding agent through the `mo-qa` skill
- Hosted browsers and Android emulators, so mobile tests need no local Android SDK or Android Studio

## Getting started

The `qa` CLI is the quickest way in. Commands below come from the momentic-ai/qa README; it needs Node.js 22.12+ or 24+ and a Momentic account.

1. Install the CLI

   Install it globally with npm (or run it through `npx`), or use the install script.

```bash
npm install -g qa
# or
curl -fsSL https://cli.momentic.ai/qa | sh
```

2. Sign in

   Log in with your Momentic account. In CI you can pass `MOMENTIC_API_KEY` or `--api-key` instead.

```bash
qa login
```

3. Start a QA session

   Give Mo a task and a URL; it explores the app and reports what it finds.

```bash
qa start "test https://your-app.com"
```

4. Hand Mo to your coding agent

   Install the `mo-qa` skill so a coding agent can start Mo sessions; `qa doctor` checks install and PATH problems.

```bash
qa skills
qa doctor
```

## When to use it

- Reach for it to bug-bash a new feature or a whole app before release without writing test scripts
- Reach for it to test every pull request's preview deployment and get findings as a review
- Reach for it when you want UI regression tests in plain English that live in your repo and gate CI
- Reach for it to cover web plus native iOS and Android flows with one QA agent

## Version history

- 2026-11-02: Momentic launched Mo on Hacker News: autonomous QA agents for web, iOS and Android, new PR testing against Vercel and Cloudflare previews, and the QA Bench write-up comparing Mo with Codex, Claude Code and Devin.