/claude-db:score

Recompute and show the two 0–100 scores from a set of findings by re-running scripts/score.mjs (pure, fully reproducible — it implements references/scoring-model.md).

• If a findings JSON path is given in $ARGUMENTS, score that file: node scripts/score.mjs --findings <path> [--paradigm <p>]. The file is a JSON array of findings (or { "findings": [...] }) conforming to schema/finding.schema.json.
• Otherwise use the findings from the most recent audit this session. If no audit has run yet, tell the user to run /claude-db:audit <target> first — do not invent findings.
• The --paradigm flag selects the category profile (PROFILES in score.mjs); the default is relational. Re-scoring under a different profile is how you show, e.g., the same Mongo findings under the document profile (no foreign-key categories in the denominator).

Show both scores with bands (A ≥ 90 · B ≥ 80 · C ≥ 70 · D ≥ 60 · F < 60), the per-category breakdown (value × weight, active?), any severity-gating cap (capped:true with the uncapped computed shown), and the needs_api count as score confidence. needs_api and speculative findings never cap. Two scores, never blended.