Menu
Auto-generate RAG priming files from client documents (PDFs, HTMLs, Word files). Reads each document, extracts key knowledge, and creates structured priming files with evidence tags. Trigger on: priming, auto-prime, procesar documentos, generar RAG, extract knowledge, documentos del cliente.
| name | auto-prime |
| author | JM Labs |
| description | Auto-generate RAG priming files from client documents (PDFs, HTMLs, Word files). Reads each document, extracts key knowledge, and creates structured priming files with evidence tags. Trigger on: priming, auto-prime, procesar documentos, generar RAG, extract knowledge, documentos del cliente. |
| allowed-tools | ["Read","Write","Edit","Glob","Grep","Bash"] |
Every unprocessed document is untapped knowledge. Auto-priming transforms raw documents into structured, queryable knowledge that agents can use.
Read all unprocessed documents in the working directory, extract structured knowledge, and generate RAG priming files that agents can reference during content generation.
List all unprocessed documents (compare against calibration-digest.md to skip already-processed files).
For each document, determine type and extraction strategy:
For each document, generate a RAG priming file:
---
source: {filename}
type: {curriculum|textbook|deliverable|insert|taxonomy|draft}
domain: {curriculo|editorial|pedagogia|neurociencia|proceso}
processed_date: {ISO timestamp}
confidence: {alta|media|baja}
---
## TL;DR
{3-5 bullet summary}
## Key Entities
{Names, systems, processes extracted}
## Constraints and Requirements
{Explicit constraints found}
## Evidence
{Key quotes or data points with [METODOLOGIA] tags}
Save priming files to references/priming-rag/rag-{domain}-{source-slug}.md
Update calibration-digest.md with new priming files and recalculated level.
Report: "N documentos procesados. N priming files generados. Nivel: {NUEVO}."
| Name | Source | Required |
|---|---|---|
| Unprocessed documents | Working directory | Yes |
| calibration-digest.md | Previous calibration | No |
| Artifact | Format | Destination |
|---|---|---|
| RAG priming files | Markdown | references/priming-rag/ |
| Updated calibration-digest.md | Markdown | Working directory |
# Auto-Prime Report — {TIMESTAMP}
## Documentos Procesados: {N}
| Archivo | Tipo | Dominio | Confianza | Priming File |
|---------|------|---------|-----------|--------------|
| {filename} | curriculum | curriculo | alta | rag-curriculo-{slug}.md |
## Nivel Actualizado: {COLD|WARM|HOT|DEEP}
## Nuevos Priming Files
- references/priming-rag/rag-{domain}-{slug}.md
## Archivos Omitidos
- {filename}: {razon}
references/priming-extraction-patterns.md — Extraction strategies per document typeAdd or improve DUA (Diseno Universal de Aprendizaje) adaptations in educational content to meet the 2+ adaptations per activity requirement across 3 principles. Trigger on: adaptaciones DUA, universal design, mejorar accesibilidad, inclusion educativa, necesidades especiales, add adaptations.
Amplify a human author's draft to Scriba quality standards — add curriculum traceability, DUA adaptations, taxonomic annotations, and style corrections while preserving 100% of original content. Trigger on: amplificar borrador, amplify draft, mejorar borrador, enriquecer draft, version 1, draft humano, author draft, delegar version 1.
Audit cross-unit consistency for progressive complexity, terminological coherence, and sequential logic across multiple editorial units. Trigger on: auditoria consistencia, cross-unit audit, contradicciones terminologicas, progresion complejidad, coherencia editorial. Detects terminology contradictions, validates Bloom/Marzano progression curves, and classifies inconsistencies by severity. Essential for multi-unit textbook coherence and Gate G3 compliance.
Auto-calibrate Scriba session by detecting available documents, calculating priming level (COLD/WARM/HOT/DEEP), and restoring session state. Trigger on: calibrar, calibration, session state, detectar documentos, priming level, entorno, environment check.
Improve existing teacher guide by incorporating pedagogical review feedback, enhancing DUA adaptations, and updating rubrics. Trigger on: mejorar guia, evolucionar guia docente, actualizar DUA, enhance guide, update rubrics. Categorizes feedback into DUA/rubric/planning/methodology dimensions, applies changes preserving indicator coverage, enforces 2+ DUA adaptations per activity, and refines rubric descriptors. Prevents pedagogical regression during guide revision cycles.
Improve existing editorial unit by incorporating review feedback while preserving curriculum traceability and pedagogical structure. Trigger on: mejorar unidad, incorporar feedback, evolucionar contenido, revisar unidad, enhance unit. Categorizes feedback, applies corrections in priority order, tracks every change, and re-validates against acceptance criteria. Prevents traceability loss during revision cycles. Essential for iterative editorial quality improvement.