# A small Vastkind allocation exercise
Use only the two supplied JSON files. Prepare an allocation for each case: exactly3 Essential Reads from home_pinned=true, then exactly6 Latest items excluding those Essential IDs, each group newest first by published_at. Deduplicate repeated identical input records by key. Preserve every source URL and timestamp. If a valid complete allocation is impossible, output status blocked and a specific reason; invent no article. Write outputs only to the designated output directory. No network, Ghost or repository changes. A result is checked against deterministic acceptance criteria; fluent explanation alone is insufficient. This is a two-case task demonstration, not a model benchmark or performance comparison.