Test Maze for QA engineers

Agents propose. You approve.

Everything an agent creates in Test Maze starts as Pending. You review it, ask for changes or mark it Reviewed, and tick the workspace’s Definition of Ready. The acceptance cases you approve are the bar every agent-written change must clear, and no MCP tool can approve on your behalf.

SOUND FAMILIAR?

What goes wrong today

“The agent already tested it”

Developers ship agent-written features faster than any QA team can review them, with nothing but the agent’s own word that it works.

Test cases nobody reads

Your best thinking lives in spreadsheets and old tickets, disconnected from the code being generated today.

Coverage you cannot prove

When someone asks “did we test permissions? accessibility?”, the honest answer takes a day of digging.

WITH TEST MAZE

What changes

A review queue for agent-written specs

Review queue · Reviewed / Changes requested

Stories and cases created over MCP are flagged as agent-created and queued oldest first. With spec-based development on, work cannot start until a person marks them Reviewed.

Definition of Ready and Done

userstory.gates · userstory.suggest_checklist

Automatic checks plus a checklist only people tick. Agents may suggest items; they cannot tick them. Stories that are not Ready cannot move into progress.

Exploratory testing that leaves a trail

exploration.start · observe · to_cases

The agent explores the running app in its own browser; Test Maze verifies each step, suggests what to try next, tracks coverage and turns the journey into test cases.

Coverage by facet, not by feel

coverage.gap_for_feature

Every feature’s cases are sorted into 8 facets (happy path, edge cases, errors, permissions, accessibility, performance, browsers, unusual data) with the gaps named.

Defects with a regression test

defect.create · defect.list

A failed run case becomes a defect with a bug breakdown, an automatic regression case and, if you connect GitLab, an issue with the link kept.

SEE IT MOVE

How it works for QA engineers

Best for: QA engineers & testers

“Explore localhost:3000, find every page”

Point your agent at any running app, local or staging.

Step 1 / 8

The full loop, step by step →

JUST ASK

You talk to your agent. It picks the tools.

“Explore my app at http://localhost:3000 and find all the pages.”
“Turn that exploration into test cases.”
“What kinds of tests is the checkout feature missing?”
“The coupon case failed. Report it as a defect and file it in GitLab.”

Test Maze for the rest of your team

Ready to give your agent a verifier?

Create a free workspace, generate an MCP token and connect your coding agent in under five minutes.