菜单
Ingiere xAPI + feedback CSV de una cohorte y produce retro report Kirkpatrick L1-L4 con iteration candidates priorizados (impact × ease). Use when user asks to "course retrospective", "retro", "post-mortem cohort", "Kirkpatrick analysis", "analizar cohort", "retro del curso", "/course-retro".
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Interactive course workbook design system — turns a course's text classes into one standalone, accessible, interactive HTML explainer (a navigable document chunked by module, Articulate-Rise lesson model; stepped mode for screen-recording). Defines the curated component kit, the content-shape -> component mapping rules, the accessibility contract, and the consumer-spec (instruction-bundle-spec.yaml) consumption + graceful-degrade rules. Consumer plugins own the brand tokens and the component vocabulary via their `workbooks` spec section; this skill stays voice-neutral.
Diálogo guiado para diseñar un Path — una super-estructura SOBRE la capa cmi5-curso que agrupa cursos existentes en milestones ordenados y los envuelve en una credential OpenBadge 3.0. NO modifica ningún course.json (aditivo). Produce path.json (fuente de verdad, validado contra path.schema.json) + content/paths/<slug>/path-overview.md (frontmatter alineado con el contrato dojo-academy). Use when user asks to "create a path", "design a path", "new path", "crear un path", "diseñar un path", "nuevo path", "ruta de aprendizaje", "learning path", "armar un path", "/new-path".
Builder's Bloom's — cognitive scaffolding for course design. Use when planning learning objectives, sequencing modules, writing assessment criteria, designing challenge progression, or reviewing whether a course builds toward higher-order thinking. Activated when discussing cognitive levels, objective writing, or skill progression.
Learning Evaluation — measuring whether courses actually work. Use when designing post-course surveys, analyzing course metrics, planning retention strategies, reviewing challenge rubrics, or discussing how to measure learning outcomes beyond quiz scores. Reference skill for evaluation planning.
Systematic research methodology for deep-diving into topics, frameworks, and tech stacks before course creation. Defines source quality standards, research artifact formats, and framework extraction criteria.
Ship-First Design — backward design for builders. Use when planning a new course, structuring modules, aligning assessments to objectives, designing challenges, or reviewing whether content connects to ship milestones. Activated when discussing course planning, curriculum design, assessment alignment, or backward design.
| name | course-retro |
| version | 1.0.0 |
| description | Ingiere xAPI + feedback CSV de una cohorte y produce retro report Kirkpatrick L1-L4 con iteration candidates priorizados (impact × ease). Use when user asks to "course retrospective", "retro", "post-mortem cohort", "Kirkpatrick analysis", "analizar cohort", "retro del curso", "/course-retro". |
Análisis post-cohort: ingesta xAPI statements + feedback forms, mapea a Kirkpatrick L1-L4, prioriza qué iterar.
course-slug (required).course.json.retros/retro-cohort-{N}.md (incrementar desde el último)."¿Tenés un CSV con xAPI statements de la cohort? (ver formato esperado en
${CLAUDE_PLUGIN_ROOT}/references/adapters/xapi/csv.md).Path al archivo (o 'skip' si no tenés):"
Si provisto: leer y parsear según adapters/xapi/csv.md.
Si skip: WARN "L2 Learning no podrá ser medido sin xAPI data" y continuar.
"¿Tenés CSVs de Tally o Typeform por módulo? (uno por módulo, formato documentado en
${CLAUDE_PLUGIN_ROOT}/references/adapters/feedback/).Lista de paths (uno por línea, o 'skip'):"
Si provistos: parsear según adapters/feedback/{tally,typeform}-csv.md.
Si skip: WARN "L1 Reaction no podrá ser medido sin feedback data" y continuar.
"Kirkpatrick L3 (Behavior 30 días después) y L4 (Results / capstone tracking) requieren input manual.
L3: ¿hiciste follow-up survey 30 días post-curso? Si sí, dame los datos clave (ej. '5/8 siguen usando Riverpod en proyectos personales').
L4: ¿cuántos alumnos completaron el capstone? ¿Con qué calidad?"
Por cada level, agregar la data:
Para cada signal anómalo (rating < 3.5, pass rate < 60%, time 2x estimated, etc.), generar un iteration candidate con:
Ordenar por score descendente. Top 5 al usuario.
docs/instructional-design/courses/{slug}/retros/retro-cohort-{N}.md:
# Retro: {course title}, Cohort {N}
**Date**: {today}
**Cohort size**: {N learners}
**Course version at cohort start**: {version}
## L1 Reaction
| Módulo | Rating avg | N | Top feedback |
|---|---|---|---|
## L2 Learning
| AU | Pass rate | Score distribution | Time vs estimate |
|---|---|---|---|
## L3 Behavior (manual)
[summary]
## L4 Results (manual)
[summary]
## Top 5 Iteration Candidates (impact × ease)
1. ...
PUERTA (sola): confirmar antes de escribir el report.
${CLAUDE_PLUGIN_ROOT}/references/adapters/xapi/csv.md${CLAUDE_PLUGIN_ROOT}/references/adapters/feedback/tally-csv.md${CLAUDE_PLUGIN_ROOT}/references/adapters/feedback/typeform-csv.md${CLAUDE_PLUGIN_ROOT}/assets/skill-references/kirkpatrick-evaluation.md${CLAUDE_PLUGIN_ROOT}/references/kirkpatrick-feedback-tools.md