LLM Card ScoringCopy Anchor Link

Scores a submitted debate-evidence card across five dimensions — relevance, clarity, uniqueness, evidence quality, usability — with a deterministic heuristic, ranks every card, flags likely duplicates, and optionally requests a real Anthropic-backed qualitative assessment.

• Route: /research/cards/scoring
• Package: debate-card-search

What it showsCopy Anchor Link

A topic switcher, a "Score card" form, a "Bulk import" textarea (----delimited batches), a ranked list with per-dimension scores and a "Likely duplicate" badge, and a "Get AI assessment" button per card. Evidence Library results also carry a "Score card" action with an inline score badge.

Data flowCopy Anchor Link

state/cardScores.ts (localStorage)
  → buildPersistedCardScoreRanking()
  → panels/CardScoringPanel.tsx

"Get AI assessment":
  → requestCardScoringAiAssessment() — lib/llm-card-scoring-client.ts
      → POST /api/reason-ai

Known gapsCopy Anchor Link

Bulk import is pasted plain text, not a real file upload. AI assessments sync to your account like card scores do (see Tool Data Sync) but are still not a shared team resource — a squad member scoring the same card sees their own assessment, not a teammate's.