| name | wizard-implementacao |
| description | Guides the client through the 9-step Implementation Guide Wizard (steering committee, TPO, RACI, comms plan, training plan, ADKAR, 3 quick-wins waves) by either pointing to the standalone HTML wizard OR conducting the conversation in chat. Output is implementation-guide-inputs.json that feeds Roadmap Part 4 PDF. Use when the user asks for "wizard", "implementation guide", "guia de implementação", "preencher parte 4", "TPO", "steering committee", "RACI", "ADKAR", "quick wins". |
Skill: Implementation Guide Wizard
What this is
The platform's Implementation Guide Wizard (app/frontend/src/components/dashboard/ImplementationGuideWizard.tsx) collects 9 structured inputs that feed Part 4 of the PDF report (Implementation Guide consolidated). This kit provides 4 ways to fill the same 9 inputs (mode D is automatic from Learning Survey):
| Mode | Best for | UX |
|---|
A. HTML wizard (wizard/implementation-guide-wizard.html) | Visual users | Browser, step-by-step, saves localStorage, exports JSON |
B. JSON template (wizard/implementation-guide-inputs.template.json) | Devs / power users | Direct edit in VS Code |
| C. Chat conversation (this skill in chat mode) | Quick draft / iteration | Copilot conducts 9 questions in chat |
| D. Auto-fill from Learning Survey ⭐ | When /plano-capacitacao already ran | Auto-extract from saida/plano-capacitacao-<DATE>.md |
All four produce the same artifact: implementation-guide-inputs.json at the workspace root.
⭐ Mode D — Auto-fill from Learning Survey output (REAL implementation)
Implementation: invoke the official script that parses saida/plano-capacitacao-<DATE>.md and generates implementation-guide-inputs.json with 6 of 9 fields auto-filled:
python3 wizard/scripts/auto_fill_from_plano.py
What it auto-fills (6 of 9):
executive_steering_committee ← Champions ativos (extraído da seção 4 do plano)
communication_plan ← Calendário sugerido (seção 5)
training_plan ← Cohorts por dimensão (seção 3)
adkar_notes ← Knowledge stage usa Top 5 workshops do plano
quick_wins_w1_4 / w5_8 / w9_12 ← Calendário 90 dias
What user MUST fill manually (Learning Survey doesn't cover):
tpo (Technology Product Owner)
raci_matrix
After running auto_fill, the user can edit implementation-guide-inputs.json to fill the 2 missing fields, then run /gerar-relatorio to render PDFs with personalized Part 4.
If saida/plano-capacitacao-<DATE>.md exists (output of /plano-capacitacao), the data ALREADY contains:
| Wizard input | Data in plano-capacitacao |
|---|
executive_steering_committee | Champions Network "ativos" (3-5 pessoas com nome+email) |
tpo | (not in learning survey — keep manual or sample) |
raci_matrix | (not in learning survey — keep manual or sample) |
communication_plan | "Calendário sugerido de workshops" (audiência + canal + cadência) |
training_plan | "Cohorts sugeridos por dimensão" (audiência × formato × cadência × Champion) |
adkar_notes | Top 5 recommendations from learning survey (Knowledge stage = workshops top 5) |
quick_wins_w1_4 | First 4 weeks of "Calendário sugerido" |
quick_wins_w5_8 | Weeks 5-8 |
quick_wins_w9_12 | Weeks 9-12 + barrier removal actions |
How to invoke Mode D:
When user runs /wizard-implementacao and saida/plano-capacitacao-*.md exists, OFFER as first option:
🎓 Detectei saida/plano-capacitacao-2026-05-08.md (do Learning Survey).
Posso EXTRAIR automaticamente: Champions, training_plan, calendário, quick wins.
Você só precisa preencher: TPO + RACI Matrix (que o learning survey não cobre).
[a] Auto-fill (modo D recomendado — preenche 6 dos 9 inputs)
[b] Modo HTML / JSON / Chat (preencher tudo manualmente)
When to use this skill
- Client asked about "wizard" / "implementation guide" / "Parte 4" / "guia de implementação"
- Before
/gerar-relatorio if implementation-guide-inputs.json doesn't exist yet (or is incomplete)
- When client wants to iterate on the implementation plan without re-rendering everything
Procedure
Default behavior: offer the 3 modes
When invoked without arguments, present the 3 options and let the user choose. Reply in PT-BR:
🧙 Wizard de Guia de Implementação — 9 inputs alimentam a Parte 4 do PDF
Como prefere preencher?
📋 [A] Modo visual (HTML standalone)
Abra: wizard/implementation-guide-wizard.html no browser.
9 passos, salva no localStorage, exporta JSON ao final.
👉 Best para: experiência visual igual ao app web.
📝 [B] Modo direto (editar JSON no VS Code)
Copie wizard/implementation-guide-inputs.template.json para
a raiz do kit-cliente/, renomeie para implementation-guide-inputs.json
e preencha cada campo.
👉 Best para: devs que preferem código.
💬 [C] Modo conversa (eu te conduzo aqui no chat)
Eu te faço as 9 perguntas uma a uma. Responda livremente.
Ao final eu salvo implementation-guide-inputs.json automaticamente.
👉 Best para: rascunho rápido, iteração interativa.
Qual prefere? Digite A, B ou C.
If user picks A or B
- Confirme o caminho e termine. (Não há nada para fazer no chat.)
If user picks C (chat mode) — conduct the 9 questions
For each step, ask ONE question at a time, wait for the answer, validate (non-empty), move on.
Questions to ask (in PT-BR):
1/9 — Comitê Executivo Diretivo
Quem são os 5-8 membros do Comitê Executivo Diretivo? Inclua: Sponsor, Programa Lead, Líder Financeiro, Líder de Segurança, Change Champion. Formato livre — pode ser uma lista com "Nome — Cargo (Papel)".
2/9 — Technology Product Owner (TPO)
Quem é o Technology Product Owner / qual a estrutura do escritório? Liste: Programa Manager (nome + dedicação %), Membros (3-5 pessoas), responsabilidades-chave, autoridade de decisão (até onde aprova sem escalar).
3/9 — Matriz RACI
Quais são as 5-8 atividades principais e seus papéis RACI (Responsible / Accountable / Consulted / Informed)? Pode listar em texto livre ou tabela.
4/9 — Plano de Comunicação
Como vão comunicar a transformação para cada audiência? Para cada uma (Toda Eng, Working Groups, Steering, Champions, Stakeholders externos), me dê: canal, frequência e owner.
5/9 — Plano de Treinamento
Qual o plano de capacitação? Para cada audiência (Devs, Champions, Eng Managers, SREs etc.), liste: formato, cadência, critério de conclusão.
6/9 — Gestão de Mudança ADKAR
Para cada estágio ADKAR (Awareness, Desire, Knowledge, Ability, Reinforcement), liste 1-3 intervenções concretas que vocês vão fazer.
7/9 — Quick Wins · Semanas 1-4
4-6 quick wins do primeiro mês — iniciativas pequenas com resultado visível em 30 dias. Indique semana sugerida.
8/9 — Quick Wins · Semanas 5-8
4-6 quick wins da segunda onda — expansão dos pilotos. Semana sugerida.
9/9 — Quick Wins · Semanas 9-12
4-6 quick wins da terceira onda — solidificando antes de escalar para H2.
After the 9 answers (chat mode)
- Show a summary of all 9 fields collected
- Ask: "Tudo certo? Posso salvar como
implementation-guide-inputs.json?"
- If yes:
- Build the JSON with the schema below
- Write to
implementation-guide-inputs.json at workspace root
- Confirm:
✓ Salvo em implementation-guide-inputs.json (X% completo). Próximo: /gerar-relatorio
Schema (output implementation-guide-inputs.json)
{
"metadata": {
"generated_at": "ISO-8601 UTC",
"generator": "wizard-implementacao skill v1.0 (chat mode)",
"completion_pct": 100
},
"implementation_guide_inputs": {
"executive_steering_committee": "...",
"tpo": "...",
"raci_matrix": "...",
"communication_plan": "...",
"training_plan": "...",
"adkar_notes": "...",
"quick_wins_w1_4": "...",
"quick_wins_w5_8": "...",
"quick_wins_w9_12": "..."
}
}
The 9 keys mirror exactly sample_payload.json::implementation_guide_inputs and the wizard React component schema.
Constraints
- DO NOT invent answers if the user gives short/vague input — ask for more detail or accept and mark as "minimal".
- DO allow markdown (tables, lists, bold) in answers — they render in the PDF.
- DO validate at the end: if any field is empty, warn but allow saving (placeholders kick in at render time).
- NEVER modify
framework.json, respostas.json, or files in relatorios/templates/.
- Output JSON to workspace root (NOT to
saida/) so the file persists across pipeline runs and serves as input, not output.
Integration with other skills
- After this skill runs,
/gerar-relatorio will automatically detect implementation-guide-inputs.json at the root and merge it into the payload before rendering the 5 PDFs.
- If client edits
implementation-guide-inputs.json manually later (e.g., updates RACI), they just rerun /gerar-relatorio — no need to redo the wizard.
Reference example
The folder wizard/ contains:
implementation-guide-wizard.html — full HTML wizard (mode A)
implementation-guide-inputs.template.json — pre-filled JSON template with rich placeholders (mode B starting point)
Both produce JSONs with the same schema described above.