Rank decarbonisation initiatives from an organisation's own emissions mix -- no questions asked.
Selects the decarbonisation initiatives for an organisation's Carbon Reduction Plan by joining its Scope 1 and 2 emissions categories against a bundled corpus of initiative templates. Runs select_initiatives.py -- no vector search, no questions -- reading crp-data.json and writing crp-initiatives.json for downstream rendering. Which initiatives appear is deterministic: the category join, then the tonnage of the category each initiative targets, then what the measure is for (abatement, then measurement, then governance), with catch-all templates last. Within one category and one kind a model orders the candidates by which most directly acts on that emission source, because every candidate there shares the same tonnage and cost is anti-correlated with substance; it can only reorder what the join produced, any failure falls back to cost and time per category, and rankedCategories records which categories the model ordered. Never states or implies how much carbon an initiative would save.