Talent Systems — Science Team
Configuration Guide

Scorer Prompt Page

Step-by-step guide for using the scorer prompt configuration page.

URL: /admin/rubrics

Page Layout

Same structure as the interviewer prompt page: left fields, right preview, sticky header, fixed save bar.

Editable Fields

System Identity

Template with {role} token. Defines how the scorer identifies its purpose.

6 Methodology Fields

These are the most important fields you'll edit:

  1. Scoring Philosophy — overall anchoring
  2. Ownership Rule — how "we" vs "I" language is scored
  3. Specificity Rule — penalty for vague answers
  4. Quantification Bonus — boost for metric-backed answers
  5. Coached Response Flag — criteria for flagging rehearsed answers
  6. Adverse Impact Guidance — when to escalate to human review

Threshold Numbers

  • Advance threshold — weighted score ≥ this → "Advance" recommendation
  • Hold threshold — weighted score ≥ this → "Hold", else "Decline"

Scoring Rules (Add/Remove List)

Additional rules appended to the prompt. Add rules for specific scoring behaviors.

Dynamic Info Cards

Some sections show info cards that can't be edited on this page:

  • Competency rubrics — shows which competencies exist (edit via Competencies page)
  • Integrity flags — shows which flags are configured (edit via Integrity page)
  • Transcript — shows that the transcript is injected dynamically

These info cards help you understand the full prompt context without switching pages.

The sticky header shows scorer-specific trends:

  • Average score across all interviews
  • Adverse impact flag rate
  • Recommendation breakdown (% Advance / Hold / Decline)
  • Delta indicators comparing to previous version

On this page