Overview
A result records what happened during execution. The runner identifies the test’s status; the detailed conversation shows the outcomes of its checks and simulation segments.
When to use it
Investigate results that differ from the expected behavior.
Use the detailed result when a response, route, tool call, or simulated outcome differs from what you expected. A canceled run should not be treated as evidence that the expected behavior passed.
How it works
Inspect the status, checks, simulation results and execution evidence.
Compare the saved expectation with the actual evidence. A response can sound appropriate while a routing or tool check still fails. Credits-limit notifications can also appear during execution; consult credits and costs when a run cannot proceed.
Example
Trace a missing tool call before changing the expectation.
If an order lookup tool was expected but never called, inspect the conversation and logs before changing the assertion. The agent may have needed more user information, or the starting variables may differ from the intended scenario. A retry should confirm the original expectation after a deliberate correction.