Three test modes
- Single fixture
- Re-run history
- Demo scenarios
Paste one input payload. The engine runs the graph end-to-end and shows per-node output in the right pane. No decision is committed.
What the test pane shows
Per node:- Status —
completed/skipped/errored/timed_out. - Elapsed — milliseconds.
- Input — the run-context slice the node consumed.
- Output — what it wrote back.
- Path — sequence of node ids the engine traversed.
- Terminal output — the final state.
- Decision payload — what would have been written if this weren’t a test run.
Mocking vendor responses
For workflows that talk to expensive or slow vendors (Crystal, BigDataCorp), the test pane supports amocks map: