Quick Start
1 — Score a user story
1
Open a work item
Navigate to any User Story in your ADO backlog. Click the TestForge tab. The heuristic INVEST score appears immediately — offline, no LLM required.
2
Read the score
- ≥ 70 — Ready: story can enter the sprint
- 40–69 — Needs work: one or more criteria are weak
- < 40 — Critical: story needs significant rework
3
Run the LLM analysis (optional)
Click Analyse (INVEST) for a deep analysis with justifications and actionable suggestions per criterion.
2 — Improve the user story
1
Choose your AC format
Toggle Bullet list or Gherkin per AC above the Analyse button.
2
Improve and review
Click Improve the US. A side-by-side diff shows exactly what changed before anything is saved.
3
Write back to ADO
Click Write back to ADO to update the work item instantly.
3 — Generate manual tests
1
Open the QA Hub
Go to Boards → TestForge in the project sidebar.
2
Set typologies and count
Select Nominal, Boundary, Equivalence, and/or Negative. Use the slider to set the total test count (2–32).
3
Generate and push
Select a user story, click Generate test cases, review the results, then click Push to ADO. No Test Plans licence required.
4 — Generate automated test code
1
Switch to Automated Tests tab
In the QA Hub, click Automated Tests.
2
Choose a framework
Playwright TS/Python/.NET, Selenium Java, Cypress JS, or Cucumber (JS/Java/Python).
3
Generate, review quality score, push
Click Generate code. Review the 5-dimension quality score. Click Regenerate if needed — warnings are automatically sent to the LLM. Then push to Repos.
You’ve completed the full TestForge workflow: score → improve → manual tests → automated test code.