Sampling and indexing
This skill standardizes how a sampling stride is chosen, how sample_ids are generated, and how interval keys and per-frame artifact stores are kept in agreement with the video's true frame range. It is the first skill consulted in the pipeline and its output feeds every downstream skill.
Scope
The skill covers stride selection, sample_id generation, interval-key formatting, per-frame artifact frame counts, and the preflight bookkeeping that keeps successive runs from contaminating each other's indices.
Where the actionable content lives
The step-by-step procedure, pseudocode, and self-check list for this skill
live in sampling-and-indexing.pdf in this directory. Open and read that
PDF before taking any sampling-related action - the Markdown file is only a
pointer and does not contain the executable workflow.