generate-rubric

Inputs

Outputs

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)