Introduction
Agentiqa is a quality gate for your app and the AI agents in it. It tests real flows, saves them as replayable plans, and blocks a release when one fails.
Agentiqa is an independent quality gate for teams shipping vertical AI agents and the apps around them. Point it at your app. An AI agent explores it in a real browser, proposes what to test, and reports what's broken with screenshots and reproduction steps. Save what matters as test plans. In CI, those plans become gates: if one fails, the release stops.
You don't write or maintain test code. You review what the agent proposes and keep what's useful.
How it works
- Explore. Give Agentiqa a URL and a goal ("test the signup flow"). The agent maps the app and ranks areas by risk.
- Curate. You approve or redirect at three checkpoints: scope, plan and findings. Only the findings always need your review.
- Save. A good exploration becomes a saved test plan (
tp_…) you can replay any time. - Gate.
agentiqa runreplays your plans and exits with a fixed code (0 pass, 1 fail, 2 config or plan limit, 3 infra). Wire it into CI and a red result blocks the release.
In headless contexts (CLI, CI, scheduled runs) every checkpoint is approved automatically. It's the same engine with no human in the loop.
Where you run it
| Surface | Use it for |
|---|---|
| CLI | The fastest start on Community Edition, local runs against localhost, and scripting |
| Coding agent | Drafting and running plans from Claude Code, Cursor or Codex |
| GitHub Action | Gating every release on your saved plans |
| Web app / Desktop app | Exploring visually and managing plans, runs and issues |
Agentiqa tests web apps. Mobile app testing isn't publicly available. It's offered to businesses on request (hi@agentiqa.com).
Next steps
- New here? Run the Quickstart. Your first gate takes about 7 minutes.
- Learn the vocabulary in Concepts.
- What's included free: Plans & limits.
- For agents:
/llms.txt,/llms-full.txt, and a.mdversion of every page.