| name | assessment-designer |
| description | Create standards-aligned K-12 assessments and scoring tools. Use this whenever a teacher or coach wants a quiz, test, formative check, performance task, item bank, or a rubric / scoring guide — including phrasings like 'write a quiz on…', 'make a rubric for…', 'I need a unit test for…', 'create a performance task…', or 'how should I grade this project?'. It aligns every item to a real, cited standard (CCSS/NGSS/state), balances difficulty and cognitive level, supplies an answer key or scoring guide, and builds in accessibility + accommodations, then self-checks against the Quality Gates and hands to quality-review. Make sure to use this whenever the user wants to measure or grade learning. Do NOT use it to build a full lesson (use lesson-planner) or slides (use presentation-builder), though an assessment may accompany a lesson. |
assessment-designer
Produces standards-aligned assessments and rubrics that actually measure the intended learning.
1. Follow the pipeline (references/method.md)
Within Generation: Analysis → Standards Alignment → Differentiation → Generation.
- Analysis — determine the artifact (
references/artifact-types.md), the objective/standard
to measure, grade band, and purpose (formative vs. summative). Log assumptions
(protocol-layer/assumptions-protocol.md).
- Standards Alignment — every item maps to a real, cited, verified standard
(
shared/standards/, protocol-layer/standards-verification.md). Alignment is the point of an
assessment — items must measure the objective, not merely the topic (QG §26).
- Differentiation & accessibility — apply UDL (multiple ways to demonstrate learning), readable
language for the band (
shared/quality/readability-age.md), and accommodations
(shared/differentiation/accommodations-catalog.md); assess content, not language, for ELs.
- Generation — build items + an answer key or scoring guide from the templates in
assets/templates/, following references/assessment-design.md (validity, cognitive balance,
bias-free items).
2. Validate, then gate
Run the universal + assessment checks (shared/quality/verification-checklists.md), self-score
against references/quality-gates.md, then hand to quality-review.
3. Always emit the metadata block
Per protocol-layer/metadata-schema.md, including standards cited, the objective measured, and
human_review_required: true. Placeholders only — never real student data.
Artifact types: references/artifact-types.md. Templates: assets/templates/. Worked example:
examples/grade3-fractions-rubric.md.
