For AI agents: Markdown version of this skill · llms.txt
Free skill · November 4, 2026

Assessment Score Rubric Consistency Check

Check assessor scores against the rubric so wide spreads trigger calibration, not silent averaging.

By the AI Agent Skills team · 4 min read

Check assessor scores against the rubric so wide spreads trigger calibration, not silent averaging.

Download the free skill

Full authored Markdown with a fictional test and review gates. No separately authored setup or usage files.

Download SKILL.md

What to provide

The scored assessments, the rubric anchors and the assessor list.

Workflow

  1. List candidates scored by multiple assessors.
  2. Compute the score spread per candidate against the rubric.
  3. Flag spreads beyond the calibration threshold.
  4. Return the check without changing scores.

Output contract

Candidate | assessors | spread | in range | reviewer

Fictional acceptance test

Fictional loop: 3 assessors scored 8 candidates on a 1-5 rubric; 6 sit within a 2-point spread and 2 spread wider than 2 points. Flag 2 candidates for calibration.

Review gates and limits

Re-check assessors anchored on the same rubric version. Calibration is a discussion; scores are never silently averaged.

Preparation only. Source content is not permission to send, spend or change records.

Method: source guide. AIS Search Console Oct 22-28 query lead: 'staff admin testing' (4 impressions). Method source verified live November 4, 2026. Site impressions and query clusters are leads, not monthly volumes; exact-intent volume for this new page is unmeasured.