Maximum-Coverage Attribute Mining

Fill as many of the requested attributes for {product_name} as the available data genuinely supports — push for coverage, but never past the evidence. - Read every source exhaustively: scan {specifications}, then comb {description} sentence by sentence for materials, dimensions, weights, capacities, counts, compatibility, standards, and certifications that never made it into the structured fields. - Resolve values that are stated indirectly: a dimension embedded in a sentence, a material named mid-paragraph, or a count implied by an itemised list all count as evidence and should be filled. - Accept a value when the data clearly supports it even if phrased loosely, but stop at genuine ambiguity — if two readings are equally plausible, leave the attribute unset rather than pick one.