generate-rubric

name	generate-rubric
author	JM Labs
description	Generate structured evaluation rubrics with 3+ performance levels aligned to Bloom and Marzano taxonomies for K-12 educational content. Trigger on: rubrica, evaluacion, criterios desempeno, niveles de logro, assessment, grading criteria, performance levels, indicadores evaluacion, rubrica analitica, escala valorativa, instrumento evaluacion. Produces DUA-compatible rubrics with observable, measurable descriptors that do not penalize expression modality.
allowed-tools	["Read","Write","Edit","Glob","Grep","Bash"]

generate-rubric

Assessment reveals learning. A rubric makes success criteria visible and achievable for every student by defining what "basico", "intermedio", and "avanzado" look like in concrete, observable terms. Without rubrics, grading is subjective; with them, both teachers and students share a common map of what mastery means.

Purpose

Generate structured evaluation rubrics with 3+ performance levels per indicator, aligned to Bloom and Marzano taxonomy levels, for K-12 educational content. Rubrics use observable, measurable descriptors and comply with DUA Principle 3 (multiple means of action and expression) by not penalizing expression modality.

Procedure

1. Read the unit's destrezas and indicadores de logro from the Matriz A&S or the generated unit, because rubric criteria derive directly from what students are expected to demonstrate.
2. Read the taxonomy mapping from references/taxonomy-mapping.md to identify the cognitive level of each indicator, because the rubric's complexity should match the taxonomy level — a "Recordar" indicator needs recognition-level descriptors, while a "Crear" indicator needs production-level descriptors.
3. For each indicador, define 3+ performance levels:
   • Basico ("En desarrollo", score 1-2): Student attempts but shows partial or incorrect performance. Describe what partial looks like concretely (e.g., "identifies 1 of 4 elements").
   • Intermedio ("Logrado", score 3-4): Student demonstrates competency with minor gaps. Describe the expected standard (e.g., "identifies 3 of 4 elements with brief description").
   • Avanzado ("Destacado", score 5): Student exceeds expectations with autonomy and depth. Describe the enriched performance (e.g., "identifies all 4 elements and explains their function").
4. Write descriptors using observable action verbs from the taxonomy level — "identifica", "clasifica", "produce", "compara" — not subjective terms like "understands well" or "shows good knowledge."
5. Include quantity thresholds in descriptors where applicable ("3 de 4 elementos", "al menos 2 ejemplos", "sin errores de concordancia"), because measurable thresholds reduce grading subjectivity.
6. Verify DUA compliance: no descriptor should require a specific expression modality as the only option. If a criterion says "writes a paragraph," add "(or presents orally, or creates a visual representation)" to allow alternative demonstration.
7. Map each rubric row to its DZ-ref and Bloom/Marzano level for traceability.
8. Generate a scoring summary showing how rubric scores translate to qualitative levels and course grades if applicable.

Inputs

Name	Source	Required
Indicadores de logro	Matriz A&S or generated unit	Yes
Taxonomy mapping	references/taxonomy-mapping.md	Yes
DUA principles	references/priming-rag/rag-dua-inclusion.md	Recommended
Target grado	User request	Yes
Asignatura	User request or inferred from DZ codes	Yes

Outputs

Artifact	Format	Destination
Rubric tables	Markdown tables	output/rubricas-<UNIT-ID>.md
Scoring summary	Markdown	Appended to rubric file
DUA compliance checklist	Markdown checklist	Appended to rubric file

Acceptance Criteria

• Every indicador de logro has a rubric with 3+ performance levels
• All descriptors use observable action verbs (no "understands", "knows", "appreciates")
• Quantity thresholds present in at least 50% of descriptors
• No descriptor penalizes expression modality (DUA Principle 3)
• Each rubric row maps to a DZ-ref and taxonomy level
• Scoring summary explains how to calculate final grades
• Evidence tags on taxonomy-level claims

Output Format

# Rubricas de Evaluacion — Unidad [ID]: [Titulo]

## Rubrica 1: [Indicador description]
**Destreza**: [REF: DZ-LENG-5-001]
**Nivel taxonomico**: Bloom: Comprender / Marzano: Comprension

| Criterio | Basico (1-2) | Intermedio (3-4) | Avanzado (5) |
|----------|-------------|------------------|--------------|
| Identifica elementos narrativos | Identifica 1 de 4 elementos (personaje, escenario, trama, conflicto) | Identifica 2-3 de 4 elementos con descripcion parcial | Identifica los 4 elementos y explica su funcion en la narrativa |
| Secuencia eventos | Ordena menos del 50% de eventos correctamente | Ordena 50-80% de eventos en secuencia logica | Ordena 100% de eventos y justifica el orden |

**Modalidades aceptadas**: Escrita, oral, visual (diagrama, mapa de historia) [DUA P3]

---

## Rubrica 2: [Next indicador]
...

---

## Resumen de Puntuacion

| Nivel Cualitativo | Rango | Equivalencia |
|-------------------|-------|-------------|
| En desarrollo | 1-2 | Requiere refuerzo |
| Logrado | 3-4 | Cumple el estandar |
| Destacado | 5 | Supera el estandar |

## Checklist DUA

- [ ] Ningun criterio exige una unica modalidad de expresion
- [ ] Descriptores son observables y medibles
- [ ] Lenguaje de descriptores es accesible para estudiantes
- [ ] Rubrica puede ser compartida con estudiantes como guia de autoevaluacion

Error Handling

• No indicadores available: Do not invent indicators. Prompt the user to provide the Matriz A&S or generated unit, because rubrics without curriculum-aligned indicators assess the wrong things.
• Taxonomy level unclear: If a destreza's Bloom/Marzano level is ambiguous, default to the level implied by the verb in the destreza description. Tag with [INFERENCIA].
• Subjective descriptors detected: If a descriptor uses non-observable language during generation, rewrite immediately. "Good understanding" becomes "explains the concept using 2+ examples."
• Single modality detected: If a criterion prescribes only one response format, add alternatives and flag with [DUA P3] annotation.

Anti-Patterns

• Vague descriptors: "Shows understanding" tells neither teacher nor student what to look for. Specify the observable behavior and quantity — "Identifies and labels 3 of 4 story elements on a graphic organizer."
• Binary rubrics: A rubric with only "correct/incorrect" is a checklist, not a rubric. Performance levels show the spectrum of mastery and give students a growth path.
• Copy-paste across indicators: Each indicator measures a different skill. Using identical descriptors with different header text suggests the rubric was not thoughtfully designed.
• Penalizing modality: Requiring "a written essay" as the only acceptable response excludes students with dyslexia or motor difficulties. Offer oral, visual, and digital alternatives per DUA Principle 3.
• Disconnected from content: A rubric criterion that does not correspond to any activity in the unit confuses teachers. Every criterion should map to specific content and exercises.

References

• references/taxonomy-mapping.md
• references/priming-rag/rag-taxonomias-bloom-marzano.md
• references/priming-rag/rag-dua-inclusion.md
• references/rubric-design-patterns.md (local)