Sanity Library Reference Docs
    Preparing search index...
    StartBlock: { filter?: string; requirements?: StartRequirement[] } & {
        kind: StartKind;
    }

    Discovery and readiness rules for standalone starts. filter evaluates against a candidate document with $tag, $definition, and $now. It controls discovery and never gates startInstance. It cannot read $fields or caller variables.

    requirements evaluates named checks in declaration order before a fresh standalone start. All must pass. Resuming an existing start and spawning children do not rerun these checks. A GROQ requirement binds $tag, $definition, $now, and $fields, with no candidate document root. $fields contains supplied input values, including GDR reference envelopes; it excludes computed defaults and hydrated document content.

    At both sites, * scans projected instances in the engine's tag. Each row contains only definition (name), subject (GDR URI or null), and completedAt (ISO timestamp or null). Completed and aborted runs are included. Raw fields such as _type, tag, and fields are unavailable. Use SingleSubjectRequirement for one unfinished run per subject. These checks are advisory; the Content Lake remains the enforcement point.