Autonomous Discovery

RoostGPT explores your web application the way a real user would — navigating pages, interacting with forms, clicking through workflows — and identifies every testable user journey automatically. No manual test plans required.

Just provide a URL. RoostGPT handles the rest, discovering login flows, CRUD operations, multi-step processes, search features, navigation paths, and more.

Exploration Modes

Choose the depth of discovery based on your needs:

Mode	Behavior	Best For
fast	Covers primary navigation paths and top-level pages	Quick smoke test coverage, CI validation
balanced	Explores forms, modals, and common interactive elements (default)	Standard coverage for most applications
thorough	Deep interaction with every discoverable element across all pages	Maximum coverage, pre-release audits

Configure via EXPLORE_MODE in your .env file or select in the Web UI.

Focus Areas

Instead of scanning an entire application, target specific sections:

# In your .env file
EXPLORE_FOCUS_AREA=checkout flow

Accepts natural language requirements like "checkout flow", "admin panel", "user registration", or "search and filtering". RoostGPT limits exploration to the specified area while discovering all related workflows within it.

Incremental Exploration

Add new focus areas to a previous run without starting over. RoostGPT builds on everything it already knows from the previous run, preserving complete history and adding new discoveries on top.

This means you can start with the core user flows, then incrementally extend coverage to admin panels, settings pages, and edge cases — all without re-running what was already discovered.

Edge Case & Negative Testing

Enable edge case detection to go beyond happy paths:

# In your .env file
ENABLE_EDGE_EXPLORATION=true

When enabled, RoostGPT automatically:

• Discovers error paths — invalid form submissions, missing data, broken navigations, error messages
• Tests boundary values — limits and edge conditions that manual testers often miss
• Merges with standard coverage — edge-case tests are combined with happy-path tests for a non-redundant, complete suite

Enable it when you need maximum coverage, or keep it off for faster standard runs.

What Gets Discovered

Element Type	Examples
Pages & routes	All navigable URLs, single-page app views, dynamic routes
Forms	Input fields, dropdowns, checkboxes, date pickers, file uploads
Authentication flows	Login, logout, registration, password reset, MFA prompts
CRUD operations	Create, read, update, delete actions on data entities
Navigation	Menus, breadcrumbs, pagination, tabs, accordions
Dynamic content	Modals, tooltips, popovers, infinite scroll, lazy-loaded sections
Error states	Form validation messages, 404 pages, permission denied screens

Importing Existing Tests

Already have a Playwright test suite? RoostGPT can scan your existing tests to understand what is already covered, then focus new generation on the gaps — no duplicates, no wasted effort.

# In your .env file
SCAN_EXISTING=true
SCAN_TEST_DIR=/path/to/existing/tests

Next Steps

• Natural Language Tests — Supplement discovery with manually described scenarios
• Page Object Model — How discovered elements are organized into POM architecture
• Self-Healing & Learning — How the system improves with repeated runs
• Configuration Reference — All exploration-related settings