Flow setup

Flow setup prepares a repository to be tested. It works out how to install, build and start the program, what the program depends on, which interfaces it offers, and how to create the data its tests need. It reads the code, so it runs whether or not the repository reads any documents.

​

When it runs

• When you connect a repository.
• After a Document scan, or a change to a repository’s links, moves the documents of a repository that was never set up.
• When you press Run or Re-run on the Flow setup row of the repository’s Pipeline tab.

Flow setup calls a model, so the workspace needs a provider on Settings › Models. When the recipe holds and the repository reads documents, Flow setup starts Flow generation.

​

The steps

Flow setup works through these steps in order. The Pipeline row and the run’s conversation on Agent show the step in progress. Each step records a fingerprint of its inputs. A later Flow setup skips a step whose inputs have not changed. The Pipeline row says how many steps settled, or why the recipe did not hold.

​

Where the results show

• The repository’s Interfaces tab lists every screen, operation, command and entry point Flow setup mapped. Filter by surface and origin. Each row opens its own page.
• The repository’s Dependencies tab lists what the program needs you to provide. See External services.

What Flow setup produces is stored per commit and carried into the next run. It is never written into your repository.

​

When the recipe does not hold

Flow setup reports Failed with the reason, and Flow generation does not start. Open the Flow setup row to read the conversation on Agent.

​

Next steps