A score card is the output returned when content is scored against a model. It carries one row per trait plus a composite row. Each trait row packs five signals into a compact view.
A score card is a structured rendering of a scoring result. It contains one row per trait declared on the model and a final row for the composite. The card is the standard display format across the web UI, the text/plain response of the REST API, and the MCP tool output. The underlying JSON response contains the same fields.
A score card for a model with multiple traits appends a composite row below the per-trait rows. The composite row reports the single aggregate score, a confidence level (the minimum per-trait confidence), and headroom (the maximum per-trait headroom — the bottleneck trait's gap). It does not carry a tier label. The aggregation rules are defined on the composite page.
A model with one trait produces a card with one trait row. The composite row is omitted because the composite would equal the single trait's score; no new information would be conveyed.
When a trait's confidence is low, the row still renders with the numeric score but the tier label is withheld and the next-break marker and headroom are not shown. The confidence indicator makes the reason explicit.