QualityMax is an independent AI testing and quality assurance platform, positioned as 'AI writes code, and an independent system verifies what actually passes.' It is aimed at development teams, especially teams without dedicated QA engineers, and connects testing and review workflows.
In terms of purpose, QualityMax is used to generate, execute, and manage tests, covering Web, mobile, API, performance, and security testing. It emphasizes 'evidence first': it not only runs tests, but also requires inspection of assertions, results, artifacts, console logs, and execution credentials before deciding whether to merge code. Its core idea is that AI code generation and AI code testing should be mutually independent systems, avoiding the situation where the same agent both writes code and writes tests, causing tests to 'pass by construction.'
Main capabilities include: generating tests from repositories, specifications, and actual application behavior; describing intent in plain English/natural language and having QualityMax directly execute end-to-end tests; generating Playwright (JS/Python) and pytest scripts, including real assertions, authentication flows, and API endpoint validation. Self-healing capabilities can detect drift when selectors fail and propose fixes. The platform supports a single repository import to generate 30+ structured test cases, with steps, tags, and priorities, and incorporates them into CI gates. It also provides AI crawling of applications, extraction of test cases from PDF/Word/spreadsheets, a persistent Playwright framework, GitHub push and CI/CD workflows, automatic smoke testing for preview deployments, quality gates that block merges, and automatic quality scoring across functionality, security, usability, and accessibility. Security testing includes prompt injection, jailbreak, PII leakage, and bias testing for AI chatbots, with 39 types of adversarial attacks and LLM-as-a-judge scoring.
In terms of usage, teams can import repositories read-only via GitHub OAuth (officially stated as not persisting source code), generate and manage tests in the hosted platform, and configure pull request reviews and quality gates. Developers can also use local 9lives and qmax-code: keep existing Playwright suites, execute locally in a bounded manner and generate receipts, review verified locator fixes, and invoke related tools from coding agents via MCP, without a platform account. The platform can connect to coding agents such as Claude, Codex, Antigravity, and Opencode via MCP to explore application behavior, draft tests, run supported frameworks, and review selector fix proposals.
Applicable scenarios includedevelopment teams that want to keep testing independent and trustworthy under AI-assisted coding; needing a unified testing view for Web, mobile, API, performance, and security; wanting to set quality gates in the PR process; and quickly generating executable tests in natural language. The platform supports self-hosted AI or bring your own key (BYO), and it is officially stated to route among Claude, GPT, and Gemini to improve resilience.
Regarding limitations, the official text clearly states: passing tests is only evidence of the behavior it checks, and does not prove that every part of the change is correct; the public code patch verification examples show recorded results for fixed synthetic examples, do not run user code in a browser, and do not prove the correctness of the entire patch; the agent and fix sections replayed in the 9lives interactive demo are real executions on a synthetic local application, do not run user code, and do not establish complete coverage. In addition, some feature details (such as pricing and specific enterprise deployment details) are not yet fully explained in the provided materials.
