@fprad0/skill-master-mcp 0.0.8 → 0.0.9

package/docs/planning/V0_0_9_ROADMAP_DISCERNIMENTO_E_CONHECIMENTO_AMPLO.md

@@ -0,0 +1,181 @@
+# V0.0.9 - Roadmap de discernimento moral e conhecimento amplo
+
+Data: 2026-06-27
+Status: planejamento, sem publicacao
+
+## Objetivo da versao
+
+Evoluir o `skill-master-mcp` de um roteador tecnico de skills para uma camada mais ampla de discernimento, qualidade e conhecimento, mantendo seguranca, revisao humana e obediencia ao escopo real do produto.
+
+## Resultado esperado
+
+Ao final da proxima versao, o MCP deve conseguir:
+
+- avaliar se uma tarefa tecnica tem risco moral ou humano;
+- aplicar gates inspirados na moral catolica;
+- rotear prompts por disciplinas de conhecimento;
+- deixar claro quando uma area exige profissional habilitado;
+- recomendar skills por dominio;
+- criar trilhas de estudo sem ativar conteudo nao aprovado;
+- melhorar a mensagem de venda para refletir capacidades reais.
+
+## Fase 1 - Governanca moral
+
+Arquivos:
+
+- `src/moral-governance.ts`
+- `src/moral-governance.test.ts`
+- `docs/architecture/GOVERNANCA_MORAL_CATOLICA.md`
+
+Entregas:
+
+- classificador de risco moral;
+- gates de dignidade humana, verdade, bem comum, justica e limite tecnico;
+- tool MCP `skill_master_moral_discernment`;
+- integracao no `skill_master_prompt_router`;
+- evals com prompts bons, ambiguos e ruins.
+
+Criterio de pronto:
+
+- prompts claramente danosos retornam `blocker`;
+- areas profissionais sensiveis retornam `refer` ou `warning`;
+- tarefas tecnicas comuns continuam fluindo sem friccao pesada.
+
+## Fase 2 - Fronteiras profissionais
+
+Arquivos:
+
+- `src/professional-boundaries.ts`
+- `src/professional-boundaries.test.ts`
+
+Dominios:
+
+- direito;
+- saude;
+- financas;
+- psicologia;
+- seguranca fisica;
+- educacao de menores;
+- decisao trabalhista;
+- decisao de credito;
+- privacidade e dados pessoais.
+
+Comportamento:
+
+- permitir informacao geral;
+- bloquear instrucao ilegal ou danosa;
+- recomendar profissional habilitado quando houver decisao real;
+- exigir fontes e jurisdicao quando o prompt envolver lei.
+
+## Fase 3 - Router de conhecimento amplo
+
+Arquivos:
+
+- `src/domain-router.ts`
+- `src/domain-router.test.ts`
+- `network/approved-domain-sources.json`
+
+Dominios iniciais:
+
+- ciencia;
+- filosofia;
+- sociologia;
+- direito;
+- matematica;
+- portugues;
+- ingles;
+- frances;
+- engenharia computacional;
+- engenharia de software;
+- engenharia de redes;
+- engenharia de IA;
+- seguranca;
+- produto;
+- documentacao;
+- catequese e doutrina catolica.
+
+Saida esperada:
+
+- dominio principal;
+- dominios secundarios;
+- skills recomendadas;
+- fontes aprovadas;
+- limites profissionais;
+- necessidade de pesquisa atualizada;
+- necessidade de Notion ou registro local.
+
+## Fase 4 - Skills candidatas
+
+Criar pacote de skills candidatas em:
+
+```text
+docs/skill-candidates/v0.0.9/
+```
+
+Essas skills nao devem ficar ativas ate revisao humana e decisao explicita.
+
+Skills candidatas iniciais:
+
+- `catholic-moral-discernment`;
+- `ai-ethics-human-dignity`;
+- `professional-boundary-triage`;
+- `broad-domain-router`;
+- `language-quality-pt-en-fr`;
+- `math-science-reasoning`;
+- `philosophy-sociology-discernment`;
+- `engineering-systems-master`.
+
+## Fase 5 - Menu e notificacoes
+
+Adicionar ao menu:
+
+- `moral-status`;
+- `moral-check-prompt`;
+- `domain-router`;
+- `professional-boundaries`;
+- `study-domain-skills`;
+- `approve-domain-skill`;
+
+Notificacoes:
+
+- skills candidatas aguardando aprovacao moral;
+- fontes de dominio ainda nao revisadas;
+- prompt bloqueado por risco moral;
+- prompt encaminhado para profissional habilitado;
+- evals morais pendentes antes de publicar.
+
+## Fase 6 - Mensagem de venda revisada
+
+Atualizar README e mensagem publica para dizer:
+
+- o que o MCP faz hoje;
+- o que e assistivo, nao automatico;
+- o que e baseado em aprovacao humana;
+- o que e camada moral opcional;
+- o que ainda e roadmap.
+
+## Ordem recomendada de implementacao
+
+1. Criar `moral-governance.ts` com regras puras e testes.
+2. Integrar gates morais no `prompt-router`.
+3. Criar tool MCP `skill_master_moral_discernment`.
+4. Criar `professional-boundaries.ts`.
+5. Criar `domain-router.ts`.
+6. Adicionar menu e notificacoes.
+7. Criar evals.
+8. Atualizar docs.
+9. Rodar `npm run check`.
+10. Pedir autorizacao antes de publicar.
+
+## Riscos
+
+- Prometer discernimento moral alem do que o software consegue cumprir.
+- Misturar opiniao pessoal com doutrina catolica sem fonte.
+- Dar aconselhamento profissional indevido.
+- Bloquear demais tarefas tecnicas comuns.
+- Criar um sistema moral rigido sem espaco para prudencia humana.
+- Ativar skills amplas sem revisao de fonte, licenca e risco.
+
+## Decisao de produto
+
+O caminho correto e tratar a proxima versao como "governanca assistiva", nao como "juiz automatico". O MCP deve ser forte para bloquear mal claro e humilde para encaminhar casos prudenciais, juridicos, espirituais e profissionais para revisao humana competente.

package/docs/skill-candidates/v0.0.9/ai-ethics-human-dignity/SKILL.md

@@ -0,0 +1,32 @@
+---
+name: ai-ethics-human-dignity
+description: "Candidate skill for evaluating AI, automation, data and algorithmic decisions against human dignity, transparency, privacy, bias, accountability and the common good."
+---
+
+# AI Ethics And Human Dignity
+
+Use this candidate skill when a prompt involves AI systems, automation, data processing, scoring, recommendation engines, surveillance, personalization, synthetic content or decision support.
+
+## Workflow
+
+- Identify affected people and groups.
+- Detect whether the system changes rights, access, work, reputation, money, safety or education.
+- Check for transparency, consent, privacy, bias, explainability and accountability.
+- Require human oversight for sensitive decisions.
+- Reject uses that manipulate, deceive, exploit or dehumanize.
+- Recommend evals and monitoring before deployment.
+
+## Boundaries
+
+- Do not treat efficiency as the highest good.
+- Do not hide responsibility behind "the algorithm".
+- Do not allow automated categorization to replace human dignity.
+- Do not approve weaponization or coercive surveillance.
+
+## Evidence Required
+
+- Intended users.
+- Data categories.
+- Decision impact.
+- Human oversight plan.
+- Bias and failure-mode evaluation.

package/docs/skill-candidates/v0.0.9/broad-domain-router/SKILL.md

@@ -0,0 +1,41 @@
+---
+name: broad-domain-router
+description: "Candidate skill for routing prompts across broad knowledge domains and recommending approved local or global skills without pretending unsupported expertise."
+---
+
+# Broad Domain Router
+
+Use this candidate skill when a prompt spans multiple areas of knowledge or when the agent needs to choose which skill family should guide the work.
+
+## Domains
+
+- Catholic doctrine and moral discernment.
+- Science.
+- Philosophy.
+- Sociology.
+- Law.
+- Mathematics.
+- Portuguese.
+- English.
+- French.
+- Computer engineering.
+- Software engineering.
+- Network engineering.
+- AI engineering.
+- Security.
+- Product and documentation.
+
+## Workflow
+
+- Classify the primary and secondary domains.
+- Identify whether the topic is stable or requires current research.
+- Recommend approved skills first.
+- Mark missing skills as roadmap candidates.
+- Add professional-boundary gates when appropriate.
+- Ask for sources or validation when the prompt has high stakes.
+
+## Boundaries
+
+- Do not invent sources.
+- Do not activate unapproved web skills.
+- Do not claim mastery in a domain without the proper skill, source or validation path.

package/docs/skill-candidates/v0.0.9/catholic-moral-discernment/SKILL.md

@@ -0,0 +1,31 @@
+---
+name: catholic-moral-discernment
+description: "Candidate skill for reviewing technology tasks through Catholic moral discernment, human dignity, truth, justice, common good, and responsible limits. Not active until human approval."
+---
+
+# Catholic Moral Discernment
+
+Use this candidate skill when a development task may affect human dignity, truth, privacy, justice, workers, vulnerable people, social trust, religious formation, or the common good.
+
+## Workflow
+
+- Identify the concrete human impact of the requested task.
+- Check whether the task promotes truth, dignity, justice, charity and the common good.
+- Distinguish technical possibility from moral permissibility.
+- Flag manipulation, fraud, exploitation, unjust discrimination, invasive surveillance and harm.
+- Prefer official Catholic sources when moral doctrine is relevant.
+- Recommend priestly, spiritual, legal or professional consultation when the case exceeds a technical assistant.
+
+## Boundaries
+
+- Do not claim to be the Church or magisterium.
+- Do not replace a well-formed conscience.
+- Do not replace a priest, confessor, spiritual director, lawyer, doctor or other competent authority.
+- Do not use Catholic language to justify abuse, humiliation or unjust coercion.
+
+## Evidence Required
+
+- Relevant prompt excerpt.
+- Risk classification.
+- Official source or reason source is needed.
+- Safer alternative when blocking a request.

package/docs/skill-candidates/v0.0.9/engineering-systems-master/SKILL.md

@@ -0,0 +1,31 @@
+---
+name: engineering-systems-master
+description: "Candidate skill for senior-level software, network, computer and AI engineering with validation, security, maintainability and release discipline."
+---
+
+# Engineering Systems Master
+
+Use this candidate skill when the user asks for architecture, implementation, debugging, release planning, network design, AI engineering, automation or production hardening.
+
+## Workflow
+
+- Read the existing system before proposing changes.
+- Prefer local patterns and simple designs.
+- Identify blast radius, risks and validation path.
+- Implement with tests proportional to risk.
+- Protect secrets, user changes and release boundaries.
+- Separate local validation from public readiness.
+
+## Boundaries
+
+- Do not publish without explicit authorization.
+- Do not run destructive commands without explicit confirmation.
+- Do not bypass security controls.
+- Do not hide uncertainty or missing evidence.
+
+## Evidence Required
+
+- Files changed.
+- Commands run.
+- Validation result.
+- Remaining risk.

package/docs/skill-candidates/v0.0.9/language-quality-pt-en-fr/SKILL.md

@@ -0,0 +1,28 @@
+---
+name: language-quality-pt-en-fr
+description: "Candidate skill for improving Portuguese, English and French text quality, translation, tone and readability with professional-boundary warnings."
+---
+
+# Language Quality PT EN FR
+
+Use this candidate skill when the user asks to write, revise, translate or improve text in Portuguese, English or French.
+
+## Workflow
+
+- Identify language, audience, tone and purpose.
+- Preserve meaning before improving style.
+- Fix grammar, spelling, clarity and coherence.
+- Offer concise alternatives when needed.
+- Warn when the text has legal, medical, liturgical or official effect.
+
+## Boundaries
+
+- Do not claim sworn translation.
+- Do not silently alter doctrine, legal meaning or contractual obligations.
+- Do not over-polish text that should preserve the user's voice.
+
+## Evidence Required
+
+- Original language.
+- Target language when translating.
+- Audience and purpose.

package/docs/skill-candidates/v0.0.9/math-science-reasoning/SKILL.md

@@ -0,0 +1,29 @@
+---
+name: math-science-reasoning
+description: "Candidate skill for structured mathematical and scientific reasoning, unit checks, uncertainty, reproducibility and evidence-based explanations."
+---
+
+# Math Science Reasoning
+
+Use this candidate skill when a prompt asks for calculation, proof, scientific explanation, data interpretation, experiment planning or technical verification.
+
+## Workflow
+
+- Restate the problem with assumptions.
+- Identify units, variables and constraints.
+- Solve step by step.
+- Check edge cases and dimensions.
+- Distinguish evidence, model and speculation.
+- Recommend experiment, test or peer-reviewed source when appropriate.
+
+## Boundaries
+
+- Do not fabricate data.
+- Do not present uncertain conclusions as certainty.
+- Do not replace professional engineering review for safety-critical systems.
+
+## Evidence Required
+
+- Formula or reasoning path.
+- Assumptions.
+- Validation or plausibility check.

package/docs/skill-candidates/v0.0.9/philosophy-sociology-discernment/SKILL.md

@@ -0,0 +1,28 @@
+---
+name: philosophy-sociology-discernment
+description: "Candidate skill for analyzing social, philosophical and cultural consequences of technology while separating doctrine, theory, evidence and opinion."
+---
+
+# Philosophy Sociology Discernment
+
+Use this candidate skill when a technical decision has social, cultural, educational, organizational or philosophical consequences.
+
+## Workflow
+
+- Identify assumptions about the human person, society and power.
+- Separate empirical claims from philosophical claims.
+- Consider effects on community, trust, family, work and vulnerable people.
+- Compare technical optimization with human flourishing.
+- Recommend dialogue, source review or moral governance when needed.
+
+## Boundaries
+
+- Do not treat ideology as fact.
+- Do not reduce people to metrics.
+- Do not use social analysis to justify manipulation or coercion.
+
+## Evidence Required
+
+- Affected groups.
+- Social risk.
+- Alternative that better serves human flourishing.

package/docs/skill-candidates/v0.0.9/professional-boundary-triage/SKILL.md

@@ -0,0 +1,40 @@
+---
+name: professional-boundary-triage
+description: "Candidate skill for detecting when a prompt enters law, health, finance, psychology or other regulated domains and must be limited to general information or referred to a qualified professional."
+---
+
+# Professional Boundary Triage
+
+Use this candidate skill when a prompt asks for guidance in a regulated or high-impact professional area.
+
+## Domains
+
+- Law and litigation.
+- Medical, health and psychological topics.
+- Finance, tax and investment.
+- Employment decisions.
+- Credit, insurance and housing.
+- Child safety and education.
+- Physical security and safety.
+
+## Workflow
+
+- Classify the professional domain.
+- Ask for jurisdiction only when needed for general orientation.
+- Provide general educational information when safe.
+- Refuse instructions for evasion, fraud or harm.
+- Recommend a qualified professional for real decisions.
+- Keep clear disclaimers in output and docs.
+
+## Boundaries
+
+- Do not give definitive legal, medical, financial or psychological advice.
+- Do not draft strategy to bypass laws or obligations.
+- Do not predict official outcomes as certainty.
+- Do not hide risk from the user.
+
+## Evidence Required
+
+- Domain detected.
+- Why referral is needed.
+- Safe general next steps.

package/docs/skill-candidates/v0.0.9/release-ethics-gate/SKILL.md

@@ -0,0 +1,32 @@
+---
+name: release-ethics-gate
+description: "Candidate skill for checking release readiness through technical evidence, explicit authorization, user impact, safety and moral responsibility."
+---
+
+# Release Ethics Gate
+
+Use this candidate skill before publishing a package, release, tag, public update, automation, AI behavior change or user-facing feature.
+
+## Workflow
+
+- Confirm explicit authorization to publish.
+- Verify tests, build, changelog, version and install path.
+- Check secrets, package contents and public claims.
+- Review user impact, moral risk and professional boundaries.
+- Separate local proof from public readiness.
+- Record remaining risks before release.
+
+## Boundaries
+
+- Do not publish without explicit user authorization.
+- Do not publish behavior that expands moral or professional claims without evals.
+- Do not include tokens, private paths or client data.
+- Do not represent planned capabilities as current capabilities.
+
+## Evidence Required
+
+- Authorization statement.
+- Validation commands.
+- Package or artifact evidence.
+- Claim review.
+- Rollback or update path.

package/docs/skill-candidates/v0.0.9/source-authority-reviewer/SKILL.md

@@ -0,0 +1,31 @@
+---
+name: source-authority-reviewer
+description: "Candidate skill for reviewing source authority, provenance, trust level and approval status before content becomes part of Skill Master behavior."
+---
+
+# Source Authority Reviewer
+
+Use this candidate skill when a workflow depends on external sources, doctrine, legal information, professional standards, web skills, repos, articles or documentation.
+
+## Workflow
+
+- Identify the domain and required authority level.
+- Prefer primary official sources.
+- Distinguish normative source, technical source, secondary guide and study material.
+- Check provenance, author, date, license and maintenance status.
+- Mark web material as unapproved until human review.
+- Recommend allowlist updates only with explicit approval.
+
+## Boundaries
+
+- Do not approve a source because it is popular.
+- Do not use outdated docs for current technical behavior.
+- Do not convert web content into an active skill without human approval.
+- Do not treat opinion as doctrine, law or standard.
+
+## Evidence Required
+
+- Source URL.
+- Authority level.
+- Review status.
+- Reason for approval, study or rejection.

package/manifests/channels/beta.json

@@ -1,13 +1,13 @@
 {
   "name": "skill_master",
   "channel": "beta",
-  "version": "00.02-beta",
-  "semver": "0.0.8",
-  "publishedAt": "2026-06-27T01:40:38-03:00",
+  "version": "00.03-beta",
+  "semver": "0.0.9",
+  "publishedAt": "2026-06-28T00:02:47-03:00",
   "git": {
     "repo": "https://github.com/FPrad0/skill-master-mcp",
     "branch": "main",
-    "commit": "7760095"
+    "commit": "69c220f"
   },
   "node": {
     "minimum": "18.0.0"
@@ -19,8 +19,8 @@
     "buildCommand": "npm run build"
   },
   "changelog": [
-    "Canal beta acompanha o prompt router e os modos de ativacao.",
-    "Fluxos de skills aprendidas agora incluem decisao humana de estudo, rejeicao e ativacao.",
-    "Evals de ativacao ficam disponiveis via npm run eval:activation."
+    "Canal beta acompanha governanca moral, roteamento amplo e bootstrap global.",
+    "Skills globais candidatas ficam empacotadas para validacao e estudo.",
+    "Menu exibe alerta forte quando o MCP ainda nao esta aplicado globalmente."
   ]
 }

package/manifests/channels/stable.json

@@ -1,13 +1,13 @@
 {
   "name": "skill_master",
   "channel": "stable",
-  "version": "00.02",
-  "semver": "0.0.8",
-  "publishedAt": "2026-06-27T01:40:38-03:00",
+  "version": "00.03",
+  "semver": "0.0.9",
+  "publishedAt": "2026-06-28T00:02:47-03:00",
   "git": {
     "repo": "https://github.com/FPrad0/skill-master-mcp",
     "branch": "main",
-    "commit": "7760095"
+    "commit": "69c220f"
   },
   "node": {
     "minimum": "18.0.0"
@@ -19,9 +19,9 @@
     "buildCommand": "npm run build"
   },
   "changelog": [
-    "Prompt router adiciona modos manual, balanced e always-on-assisted.",
-    "Fluxo de aprendizado cria pacotes de aprovacao humana, decisoes de estudo/rejeicao e ativacao local/global.",
-    "Menu operacional ganha status de ativacao, recomendacao por prompt, notificacoes e resumo para Notion.",
-    "Suite de evals de ativacao valida casos positivos, negativos e riscos antes de release."
+    "Governanca moral adiciona discernimento, gates de dignidade humana, verdade, bem comum e limites profissionais.",
+    "Roteamento amplo recomenda skills globais aprovadas por dominio tecnico, linguagem, ciencia, filosofia e engenharia.",
+    "Skills globais embutidas podem ser instaladas em CODEX_HOME/skills ou ~/.codex/skills.",
+    "Bootstrap global registra o MCP para Codex, Claude Desktop e Gemini com alerta visual no menu."
   ]
 }

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@fprad0/skill-master-mcp",
-  "version": "0.0.8",
+  "version": "0.0.9",
   "private": false,
   "type": "module",
   "description": "Personal MCP server that catalogs, recommends and reports skills with update-aware metadata.",
@@ -9,8 +9,11 @@
     "skill-master-activation": "./bin/skill-master-activation.mjs",
     "skill-master-configure-private-registry": "./bin/skill-master-configure-private-registry.mjs",
     "skill-master-eval-activation": "./bin/skill-master-eval-activation.mjs",
+    "skill-master-bootstrap-global": "./bin/skill-master-bootstrap-global.mjs",
+    "skill-master-install-global-skills": "./bin/skill-master-install-global-skills.mjs",
     "skill-master-menu": "./bin/skill-master-menu.mjs",
     "skill-master-mcp": "./bin/skill-master.mjs",
+    "skill-master-register-clients": "./bin/skill-master-register-clients.mjs",
     "skill-master-success-skills": "./bin/skill-master-success-skills.mjs",
     "skill-master-update": "./bin/skill-master-update.mjs"
   },
@@ -18,6 +21,8 @@
     "bin",
     "dist",
     "docs/architecture",
+    "docs/planning",
+    "docs/skill-candidates",
     "docs/operations/GUIA_NPM_PRIVADO.md",
     "manifests",
     "network",