Test paths
A test path is one specific narrative through a scenario. One scenario almost always has multiple realistic conversations; paths are how AssureAgent expands a single piece of authoring into many runnable tests.
How paths come into existence
You generate paths by clicking Generate test paths on a scenario. Generation behavior depends on design style:
Free-flow scenarios
The platform reads your description and produces several alternative narratives that exercise different ways the conversation could go. From a single billing-dispute description, you might get:
- Happy path — refund granted, polite end.
- Verification fails — caller can’t be looked up.
- Escalation — caller demands a supervisor after pushback.
- Edge — caller gives the wrong card number, agent has to handle gracefully.
Each path has its own name and synopsis. You can preview them before running.
Graph scenarios
Each branch combination in the graph becomes one path. A graph with two binary branches expands to four paths automatically. Service-node responses (mocked HTTP calls) are also varied across paths to cover failure modes.
Path properties
Every path owns:
- Name + synopsis — what makes this path different from its siblings.
- Persona prompt — the actual caller persona that will be used at runtime, derived from the scenario description (free-flow) or the path’s branch sequence (graph).
- Success criteria — inherited from the scenario by default, can be overridden per path.
- Status —
not run,running,passed,failed,error.
Re-generating paths
If you change the scenario description (or the graph), the existing paths don’t automatically refresh — they keep the persona they were generated with. To pull in the new authoring, click Generate test paths again.
Running paths
Click Run on a row to execute one path against the target. You can also run all paths in a scenario at once — the platform queues them and runs them with bounded parallelism. Read more →
Where to find them
Open any scenario from the Scenarios page. Paths are listed under the scenario detail; you can filter by status (failed only, not yet run) and bulk-select for batch runs.