A whole lockfile in one request (pilot findings, Priority 3): compact, comparable per-package results in submitted order, with a top-level status distinguishing full coverage, partial coverage, ambiguity and no coverage (Priority 4) instead of one flat dict a caller has to interpret differently for every failure mode. view is summary (default -- foresight_api.risk_batch's compact fields) or detail (the same full per-package answer /v1/risk gives, keyed by query). Ambiguous or unmatched entries never raise for the whole batch -- one bad entry becomes one result, not a 4xx for packages that would have resolved fine.

Every result, in both views, carries rank: an integer where 1 is the highest triage.score in this batch, ties broken by submission order; null when the entry has no score (not_covered, ambiguous, service_error, or insufficient evidence). The list itself stays in submission order -- sort by rank to get the riskiest first.