Guided Mode¶
Overview¶
Guided is one of VibeTesting's two execution types (alongside Exploration). In guided mode, you provide Tests Guidelines for each flow — either as high-level goals or as explicit step-by-step instructions — and the agent follows the intent strictly (staying on-path to complete the described journey).
When to use guided mode¶
Use guided mode when you want:
- To test a specific user flow (as a goal or as step-by-step)
- Repeatable regression checks for known journeys
- The agent to stay focused on completing your flow rather than exploring freely
How guided mode differs¶
- vs. Exploration — the agent stays on the flow you described instead of freely exploring the app.
How to write Tests Guidelines¶
Write each test as either:
- A goal (what the journey should accomplish), or
- Step-by-step instructions (what to click/type, in order)
Examples (goal-based):
Sign up as a teacher, log out, then sign in with the new details. Verify sign-in succeeds.
Add three items to the cart, proceed to checkout, and verify the total matches.
Examples (step-by-step):
1) Click "Login"
2) Enter email "user@example.com"
3) Enter password
4) Click "Sign in"
5) Verify the dashboard loads without errors
The system normalizes your Tests Guidelines (goals or steps) into flow goals and executes them as goal-based runs. Only free-roaming exploration runs without flow goals.
Tips for step-by-step Tests Guidelines¶
When you choose step-by-step guidelines, write each step in natural language with:
- The action (click / input text / scroll)
- The target element (what you want to click or type into)
- Optional input text
Tips:
- Prefer visible UI labels ("Add Item", "Email", "Save") when possible
- Keep each step to a single action
- Be specific if there are multiple similar buttons/fields
Example:
Click the "Add Item" button
Click the "Insert item name here" input and enter "My item"
Monitoring and results¶
Guided tests run in the cloud. While running:
- Monitor progress via the agent chat or real-time logs
- View step-by-step actions and detected issues when the run completes
- Optionally receive an email report with screenshots