Help
FAQ
The questions engineers ask before they trust a tool with their bench.
Yes. The package is a standalone Python application. Download it, commit it, run it on a machine with no TestFlow installed, hand it to a colleague. There is no runtime licence check inside it.
No. Runs execute on your machine, in the desktop app, over your local VISA layer. What syncs back to your workspace is the run's outcome and the report, so the Report pane and run history can show them.
No. You never have to open the Code pane. It is there because engineers want to see what will touch their hardware, and because a Python step is the right answer when a calculation is genuinely custom.
Three options, in order of how well they work:
- Try a model from the same family. Vendors usually share a command set across a series.
- Use a
Serialinstrument if it has a text command interface. - Ask in chat. The agent can flag the gap with the type, manufacturer, and model attached, and the catalog grows continuously.
Not directly, and converting them is usually the wrong goal. The faster path is to describe what the script does and let TestFlow build it, then compare results on the same DUT. You end up with something the rest of your team can read. A TestFlow script file can be imported into Executor and run as-is.
Yes, for everything except a bench run. Configure the models you expect, build the workflow, generate the plan, wire the schematic, and simulate. When the hardware arrives, add the addresses and run.
It is built not to be able to. The report structure has no field capable of holding a measured value, stating that a test passed is a gated operation that refuses without a current passing run, and a failing limit is reported rather than retried or edited away. The mechanisms are listed in what a result means.
You correct it in a sentence and it patches the step. If an edit made things worse, revert the turn or restore a revision. Nothing the agent does is irreversible, and nothing it writes reaches hardware without you starting a run.
It means the sequence is coherent. It says nothing about your hardware, because the data is synthetic. Only a bench run tells you about your part.
Yes, by sharing the project. A shared project carries its bench, so a colleague opening it sees the same instruments. Addresses are machine-specific, so they will set their own.
AI generations: building and editing workflows, writing plans, shaping reports, wiring schematics. Running tests does not, and neither does an all-SCPI workflow compiling its Python, which happens locally with no model call.
Instrument control is already on-premise: it runs on your machine and your traffic does not leave it. For a fully air-gapped deployment, talk to us about what you need.