In addition to its dense-seeded local diffusion retriever, the paper compares against an explicit budgeted best-first traversal of the prerequisite graph. This baseline is motivated by anytime search under resource constraints (hansen2001monitoring, hansen2007anytime): the traversal expands the highest-scoring frontier nodes first and can be stopped once a node, edge, or token budget is exhausted, returning the best context found so far. The role of this baseline is to test whether the paper's diffusion-based policy improves over a strong budget-respecting traversal that exploits the same prerequisite graph but uses best-first search rather than additive-penalty max-aggregated diffusion.