Explorer
Topic-based sessions that organize tests by topic and let suites evolve through outputs, evaluation, and AI suggestions.
Overview
Explorer extends static test sets by organizing tests around topics and letting suites grow over time. Instead of a fixed set of tests, Explorer helps you map coverage gaps, add tests in a structured way, and generate outputs that feed back into evaluation.
Key concepts
Topics: Tests are grouped by topics—thematic areas you want to cover. Topics act as a navigation layer for what is passing, failing, or missing.
Test explorer UI: In the Rhesis app, Explorer (under Testing) gives you a browsable tree, bulk actions, suggestions, and export to a regular test set.
Output generation: Explorer supports generating stored outputs against your endpoint, then evaluating them with selected metrics.
Workflow
- Define topics that reflect behavioral areas
- Add tests within each topic—manually, via import, or suggestions
- Generate outputs in bulk against your endpoint
- Evaluate to find gaps and failures
- Iterate with suggestion flows before persisting
SDK usage
The SDK exposes tree APIs under (TestTree, schemas). These call the same backend routes as the UI.
Difference from static test sets
Regular test sets are a fixed bundle for execution. Explorer sessions emphasize an evolving topic tree, curated outputs, and suggestion-driven growth.
Best practices
- Name topics by intent or risk area, not tiny prompt variants
- Re-run evaluation after material model or prompt changes
- Use export when you need a frozen, standard test set for CI