A position is the role you are hiring for. Hiring flows, screening and scoring all read it, so it is worth getting right before you build a flow on top.
The essentials
Title and level. The level pitches question difficulty; a senior role left unset gets mid-level questions.
Skills, with weights. This is the important part. Each skill you list becomes an axis candidates are scored on, and the weight decides how much it moves the overall score. A list of fifteen equally weighted skills produces a score that means very little.
Required vs preferred. Drives screening and how a gap is reported.
Location and working pattern. Shown on the job board and used in matching.
Defaults that cascade
A position can carry defaults that every flow built on it inherits: evaluation leniency, a scoring configuration, proctoring settings, and result visibility. Setting them here means not setting them on every stage of every flow.
They are inherited, not copied-and-frozen at creation for the runtime cascade — a stage that does not override a value resolves it from the position at the time it runs.
Publishing
Publishing puts the role on the public job board. You can create flows and interview people on an unpublished position; publishing is only about public discoverability.