Agent QA

Agent QA is a source-available AI testing agent for developers and software teams testing web, Android and iOS apps. You describe a user journey and its expected results in plain English. It tests the real interface and carries observations from earlier runs into later ones, so it can account for how your app behaves.

Its execution can recover when a click, form entry or selection fails: the agent checks the interface again and attempts another route within the same test. For web apps, it works against local or staged targets without hardcoded selectors. Memory records product behavior and test context, including successful recoveries, while a cache reuses validated action plans when the screen still matches. This can reduce repeated model work and token use.

Developers can run tests through a dashboard or CLI. Coding agents can use MCP and skills to author tests, examine recorded evidence and use the results to guide code changes. Tests and configuration live in repository files; memory and run artifacts remain under your control. Suites group related journeys, and reusable web login states support flows that require authentication.

You choose the model backend. Agent QA supports local models, OpenAI- and Anthropic-compatible endpoints, Gemini, and Codex or Claude Code subscriptions. Hosted backends handle model requests outside your machine. Optional Docker-isolated hooks can prepare test data, call APIs, check side effects and clean up after a run.