izanagi-ai 2.0.0

package/AGENTS.md

@@ -0,0 +1,53 @@
+# AGENTS.md — NexusAI
+
+NexusAI é um **framework meta** (instruções para agentes de IA), não um projeto executável. O repositório inteiro é markdown + JSON — sem npm, sem build, sem testes, sem CI.
+
+## Arquitetura
+
+- `SYSTEM.md` — Fundação: identidade, princípios, arquitetura em camadas (Decision → Context → Skill → Quality → Reflection → Memory)
+- `RULES.md` — 9 regras operacionais + formato de declaração de skills + regras de comunicação
+- `core/` — 8 engines (decision-engine, context-engine, token-manager, compression-engine, reflection-engine, evolution-engine, quality-gates, planning-engine)
+- `agents/` — 10 agentes pré-definidos como JSONs que compõem skills
+- `skills/INDEX.md` — Registro central de todas as 111 skills. **Qualquer skill nova deve ser registrada aqui**
+- `core/skill-resolver.json` — Mapeia short IDs para paths de skills. **Qualquer skill nova deve ter alias aqui** para ser usada por agentes
+
+## Regras-chave
+
+- **Toda skill precisa de entrada em dois lugares**: `skills/INDEX.md` + `core/skill-resolver.json`
+- **Formato de skill**: YAML header (name, version, priority, dependencies, triggers, token_budget, compatibility) + seções (Identity → Goals → Workflow → Decision Tree → Rules → Checklists → Metrics → Evolution)
+- **Agentes**: JSON com `skills` (array de short IDs), `chains` (mapa de tipo de task → chain de skills), `always`/`never`
+- **Task routing**: `RULES.md:86-106` — 6 classificações (new_project, bug, refactor, review, question, security_audit) + fallback genérico
+- **Output format** obrigatório (RULES.md seção 2.1): `## Context` → `## File` (com path) → `## Notes`
+- **Token budgets** (SYSTEM.md:115-120): 2048 soft / 4096 hard por resposta, 8192 contexto, compressão automática >70%
+- **Quality Gates** (SYSTEM.md:129-135): Security → Style → Clarity → Conciseness → Completeness (nesta ordem)
+- **5 proibições** (RULES.md:49-53): não adivinhar APIs, não codificar sem entender o codebase, não repetir contexto, não ignorar convenções, não hardcodear secrets
+
+## Estrutura de diretórios
+
+| Diretório | Conteúdo |
+|-----------|----------|
+| `core/` | Engines do sistema (8 skills) |
+| `architecture/` | Padrões arquiteturais (Clean Arch, Hexagonal, DDD, CQRS, etc.) |
+| `coding/` | Skills de engenharia (backend, frontend, React, Laravel, etc.) |
+| `skills/` | 77+ skills especializadas (quality, debugging, cloud, devops, etc.) |
+| `agents/` | Definições de agentes como JSON |
+| `memory/` | Gerenciamento de memória (6 skills) |
+| `teaching/` | Modo professor e ensino adaptativo |
+| `testing/` | Testes unitários, integração, E2E, mocking |
+| `security/` | OWASP, pentest, segurança |
+| `database/` | SQL, PostgreSQL, MySQL, Redis |
+| `devops/` | Docker, K8s, CI/CD, Linux |
+| `optimization/` | Redução de tokens, otimização de prompt, custo |
+
+## Convenções
+
+- **Versão atual**: v2.0.0 (SemVer). CHANGELOG.md e ROADMAP.md mostram o plano até v3.0.0
+- **Idioma**: maior parte em português (descrições de agentes, READMEs); conteúdo técnico em inglês
+- **Filosofia**: "Architecture first. Code second." — nunca pular planejamento
+- O framework é **carregado** via `load NexusAI/SYSTEM.md`, não via runtime
+
+## Design Preference (Default)
+
+- **Estilo visual**: Futurista Apple-like — fundo escuro, glassmorphism, gradientes suaves, partículas animadas, 3D tilt em cards, tipografia bold, animações sutis, sem excesso de "cara de IA"
+- **Stack**: a stack (React, Vue, HTML/CSS/JS, etc.) é definida pelo contexto do projeto ou pelo que o usuário pedir — o estilo visual se adapta à stack, não o contrário
+- **Exceptions**: quando o usuário pedir um estilo específico, seguir a solicitação

package/CHANGELOG.md

@@ -0,0 +1,198 @@
+# Changelog
+
+> All notable changes to the Nexus AI framework.
+
+---
+
+## [2.0.0] — 2026-07-22
+
+### Changed
+- Version bump to v2.0.0 (all Phases 1-6 implemented)
+- Cleansed project-specific references for public release
+- Fixed all broken paths in decision-engine skill chain matrix
+- Registered 31 orphan skills in INDEX.md and skill-resolver.json
+- Corrected count mismatches across INDEX.md, CHANGELOG.md, ROADMAP.md
+- Updated ROADMAP.md to reflect actual implementation state
+
+## [1.0.0] — 2026-07-17
+
+### Added
+
+#### Core
+- `SYSTEM.md` — Foundation: identity, principles, architecture, token budget, quality gates, memory, evolution.
+- `RULES.md` — 9 golden rules, skill declaration, communication, memory, quality, security, error recovery.
+- `README.md` — Entry point, quick start, principles table.
+- `decision-engine.md` — 15 category classification, skill chain matrix, keyword routing, 70% confidence threshold.
+- `context-engine.md` — 4-section context window, load prioritization, compression algorithm, relevance scoring.
+- `token-manager.md` — 4-tier budget, priority-based allocation, real-time monitor, compression triggers.
+- `compression-engine.md` — 4 compression levels (lossless → emergency), 5 strategies, decision preservation.
+- `reflection-engine.md` — Post-task self-review, 5-dimension scoring, pattern detection (50-task rolling window).
+- `evolution-engine.md` — 6 pattern → action mappings, 4 change types, auto-apply rules, full traceability.
+- `quality-gates.md` — 5 gates (security, style, clarity, conciseness, completeness), security is fatal.
+- `planning-engine.md` — Atomic step decomposition, topological sort, effort estimation, circular dependency detection.
+
+#### Memory (6)
+- `memory-manager.md` — 3-tier (session/project/long-term), JSON storage, knowledge graph, recall engine.
+- `session-compression.md` — 200-token session summary, decision/error/action preservation.
+- `conversation-summarizer.md` — 68:1 compression, extraction priority, YAML summary format.
+- `context-recovery.md` — Session recovery flow, recovery prompt template.
+- `smart-recall.md` — Relevance scoring (keyword 0.4 + recency 0.3 + relation 0.3).
+- `long-term-project-memory.md` — Persistent project context across sessions.
+
+#### Optimization (3)
+- `token-reducer.md` — 6 reduction techniques, format selection matrix (8 types).
+- `prompt-optimizer.md` — 4 optimization passes (noise, implicit, ambiguity, structure).
+- `cost-optimizer.md` — Token cost tracking, infrastructure optimization strategies.
+
+#### Architecture (10)
+- `clean-architecture.md` — Layers, directory structure, dependency rule.
+- `hexagonal-architecture.md` — Ports & Adapters, testability, infrastructure swap.
+- `cqrs-specialist.md` — Read/write separation, when to use/avoid.
+- `event-driven-architect.md` — Events, brokers (RabbitMQ/Kafka/Redis), idempotency.
+- `ddd-specialist.md` — Building blocks, ubiquitous language, aggregate example.
+- `microservices-expert.md` — When to use/avoid, key patterns, service template.
+- `monolith-expert.md` — Modular monolith, migration path to microservices.
+- `repository-pattern.md` — Interface contract, implementation, rules.
+- `unit-of-work.md` — Transactional consistency, commit/rollback, clear.
+
+#### Coding (13)
+- `backend-engineer.md` — Multi-language (PHP, Node, Python, C#), conventions, checklist.
+- `frontend-engineer.md` — React + TS + Tailwind, state management (5 states), a11y, performance.
+- `api-designer.md` — REST conventions, response envelope, auth decision tree, rate limiting, OpenAPI.
+- `laravel-specialist.md` — Eloquent, Form Requests, Policies, conventions, patterns.
+- `php-specialist.md` — PHP 8.x features, PSR-12, PHPStan level max config.
+- `javascript-specialist.md` — ES2022+, functional patterns, checklist.
+- `typescript-specialist.md` — Strict mode, discriminated unions, branded types, tsconfig.
+- `python-specialist.md` — Python 3.11+, typing, dataclasses, async.
+- `react-specialist.md` — Hooks, RSC, compound components, performance checklist.
+- `vue-specialist.md` — Composition API, Pinia, TypeScript, conventions.
+- `nodejs-specialist.md` — Express/Fastify, error handling, services, checklist.
+- `java-specialist.md` — Java 17+, Spring Boot 3, records, virtual threads.
+- `csharp-specialist.md` — .NET 8, Minimal APIs, primary constructors, records.
+
+#### Security (3)
+- `security-engineer.md` — OWASP Top 10, secrets detection (7 regex), security report.
+- `owasp-auditor.md` — Full OWASP Top 10 audit, CVSS severity, CVE references.
+- `pentest-reviewer.md` — 5 attack categories (IDOR, priv esc, mass assignment, SSRF), report format.
+
+#### Quality (12)
+- `senior-code-reviewer.md` — 6-dimension scoring, 4 severity levels, YAML report.
+- `clean-code-validator.md` — 6 validation categories, function size heuristic, before/after examples.
+- `solid-validator.md` — 5 principles with checklists, score, refactoring recommendations.
+- `dry-kiss-yagni-validator.md` — 3 principles with checks and examples.
+- `complexity-analyzer.md` — Cyclomatic complexity (McCabe), cognitive complexity, thresholds.
+- `bug-prevention.md` — 4 prevention layers, 8 bug patterns with detection and prevention.
+- `design-pattern-advisor.md` — Decision tree, pattern suggestions, anti-patterns.
+- `refactoring-specialist.md` — 12 code smells, 3 techniques, safety checklist, plan template.
+- `technical-debt-analyzer.md` — 6 debt categories, estimation formula, prioritized backlog.
+- `breaking-change-detector.md` — API and database breaking changes, detection flow.
+- `performance-optimizer.md` — Audit workflow, bottleneck types, caching strategy (4 levels).
+- `scalability-expert.md` — 4 scaling dimensions, horizontal scaling checklist, sharding.
+
+#### Testing (4)
+- `unit-test-engineer.md` — Pest/Jest/pytest/xUnit, Arrange-Act-Assert, edge case checklist, naming convention.
+- `integration-test-engineer.md` — Scope, Laravel/Pest example, database testing.
+- `e2e-test-engineer.md` — Cypress/Playwright, critical journeys, example.
+- `mocking-specialist.md` — 5 double types, frameworks (Mockery, Jest, Mockito, Moq), rules.
+
+#### DevOps (8)
+- `devops-engineer.md` — 8-step workflow, Docker, CI/CD, monitoring, backup, security hardening, runbook.
+- `docker-expert.md` — Multi-stage, image optimization, compose for dev, security.
+- `kubernetes-specialist.md` — Deployment template, resources (ConfigMap, Secret, Ingress, HPA).
+- `git-expert.md` — Trunk-based vs Git Flow, commit conventions, useful commands.
+- `git-flow-specialist.md` — Branch structure, workflow, commands.
+- `ci-cd-specialist.md` — Pipeline stages, GitHub Actions example, quality gates.
+- `linux-specialist.md` — Server hardening, performance tuning, commands.
+- `windows-specialist.md` — IIS, PowerShell, Windows-specific config.
+
+#### Database (6)
+- `database-engineer.md` — Naming conventions, data types, index strategy, migration safety levels.
+- `sql-optimizer.md` — EXPLAIN ANALYZE, index rules, sargable predicates, query rewrites.
+- `postgresql-specialist.md` — JSONB, full-text search, partitioning, CTEs.
+- `mysql-specialist.md` — InnoDB, EXPLAIN FORMAT=JSON, performance tuning.
+- `sqlserver-specialist.md` — T-SQL, execution plans, indexed views.
+- `redis-specialist.md` — Data structures (5 types), use cases, eviction policies.
+
+#### Engineering Roles (4)
+- `principal-engineer.md` — Technical vision, org-level standards, decision framework.
+- `staff-engineer.md` — Deep technical excellence, large feature delivery, mentorship.
+- `tech-lead.md` — Team leadership, daily rhythm, engineering-product bridge.
+- `cto-advisor.md` — Strategic advice, stakeholder communication, budget and risk.
+
+#### Debugging (3)
+- `bug-hunter.md` — 8-step protocol, binary isolation, debugging decision tree, bug report format.
+- `debug-specialist.md` — 6-step protocol, error pattern library (4 patterns), quick diagnostics CLI.
+- `root-cause-analyzer.md` — 5 Whys, Fishbone, Premortem, 7 root cause categories, pattern detection.
+
+#### Teaching (6)
+- `professor-mode.md` — 4-level detection (beginner→expert), 4 teaching strategies, interactive exercises.
+- `mentor-mode.md` — 4 guidance principles, Socratic questioning, learning roadmap generator.
+- `code-explainer.md` — 4 explanation levels (overview→expert), YAML format, pattern highlighting.
+- `interactive-teaching.md` — 6 exercise types, interaction flow with correction feedback.
+- `adaptive-teaching.md` — Difficulty/pace/style adaptation rules based on performance.
+- `learning-tracker.md` — Persistent learning record, confidence scoring, progress report.
+
+#### Analysis (5)
+- `requirement-analyzer.md` — Extraction process, functional/non-functional categorization, structured format.
+- `risk-analyzer.md` — Probability × Impact matrix, risk register, mitigation planning.
+- `dependency-analyzer.md` — Security, freshness, compatibility, license audit, report format.
+- `tradeoff-analyzer.md` — 6 criteria with weights, weighted scoring, recommendation.
+- `alternative-solution-generator.md` — 3+ options format with pros/cons/effort per option.
+
+#### Documentation (6)
+- `documentation-writer.md` — 5 documentation types, README template.
+- `technical-writer.md` — Diátaxis framework (tutorials/how-to/reference/explanation), writing principles.
+- `readme-generator.md` — Automated sections, badge generation, stack extraction.
+- `uml-generator.md` — PlantUML + Mermaid class/component/use case diagrams.
+- `sequence-diagram-builder.md` — PlantUML + Mermaid sequence diagrams with actor flow.
+- `er-diagram-builder.md` — PlantUML + Mermaid ER diagrams from schema.
+
+#### Project Management (3)
+- `project-manager.md` — Milestones, sprint tracking, velocity, stakeholder communication.
+- `task-planner.md` — Atomic task breakdown, estimation guidelines (1-8 points), acceptance criteria.
+- `release-planner.md` — Version bump rules, release checklist, changelog entry format.
+
+#### UX/Observability (5)
+- `ux-reviewer.md` — 10 Nielsen heuristics, severity-graded report.
+- `accessibility-reviewer.md` — WCAG 2.2 AA (perceivable/operable/understandable/robust), audit tools.
+- `logging-expert.md` — Structured JSON logging, what to log/not log by level.
+- `observability-expert.md` — 3 pillars (logs/metrics/traces), RED metrics, dashboard structure.
+- `monitoring-specialist.md` — Alert rules (critical/warning), incident response (6 steps).
+
+#### Self-Improvement (6)
+- `self-correction.md` — Error detection, correction protocol (acknowledge/explain/correct/prevent/log).
+- `self-critique.md` — 8 critique questions, proactive revision process.
+- `continuous-improvement.md` — Improvement cycle, effectiveness tracking.
+- `hallucination-detection.md` — 4 confidence levels (90%+ to <50%), detection patterns.
+- `confidence-estimator.md` — Source reliability scoring, communication patterns per confidence level.
+- `continuous-learning-engine.md` — 3 learning sources, knowledge gap detection.
+
+#### Skills (1)
+- `INDEX.md` — Complete registry of all 111 skills with status and file paths.
+
+---
+
+### Stats
+
+| Category | Skills |
+|----------|--------|
+| Core | 8 |
+| Architecture | 10 |
+| Coding | 13 |
+| Security | 3 |
+| Quality | 12 |
+| Testing | 4 |
+| Database | 6 |
+| DevOps | 8 |
+| Engineering Roles | 4 |
+| Debugging | 3 |
+| Teaching | 6 |
+| Memory | 6 |
+| Optimization | 3 |
+| Analysis | 5 |
+| Documentation | 6 |
+| Project Management | 3 |
+| UX/Observability | 5 |
+| Self-Improvement | 6 |
+| **Total** | **111** |
+

package/LICENSE

@@ -0,0 +1,21 @@
+MIT License
+
+Copyright (c) 2026 Pedro Henrique Sanches Leal
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT{TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.

package/README.md

@@ -0,0 +1,96 @@
+# NexusAI
+
+Framework modular, skill-oriented para agentes de IA especializados em desenvolvimento de software e automação.
+
+> **Filosofia:** Arquitetura primeiro. Código depois. Qualidade medida. Evolução contínua.
+
+---
+
+## Instalação & CLI via NPM
+
+O NexusAI possui uma **CLI executável** que pode ser utilizada via `npx` ou instalada globalmente.
+
+```bash
+# Execução direta via npx (sem instalação permanente)
+npx nexusai-cli run "Criar API REST com Laravel e JWT"
+
+# Ou instalação global
+npm install -g nexusai-cli
+
+# Agora use diretamente o comando nexus / nexusai
+nexus run "Refatorar camada de autenticação"
+nexus list skills
+nexus doctor
+```
+
+---
+
+## Comandos Principais da CLI
+
+| Comando | Descrição |
+|---|---|
+| `nexus run "<task>"` | Analisa a tarefa, seleciona o agente ideal e mapeia a corrente de skills. |
+| `nexus compile <agente> [arquivo]` | Compila um System Prompt completo do agente + fundação do sistema para arquivo ou stdout. |
+| `nexus init` | Inicializa regras `.nexus/` no seu projeto atual. |
+| `nexus list [skills\|agents]` | Lista todas as skills e agentes registrados com seus aliases. |
+| `nexus doctor` | Valida integridade do framework, JSONs de agentes e mapeamentos de aliases. |
+
+---
+
+## Estrutura do Projeto
+
+```
+NexusAI/
+├── bin/            Executável da CLI (bin/nexus.js)
+├── src/cli/        Código fonte dos comandos CLI
+├── core/           Motor central (decisão, contexto, reflexão, skill-resolver)
+├── agents/         Definições de Agentes de IA em JSON
+├── skills/         Skill base (80+ skills especializadas em Markdown)
+├── memory/         Gerenciamento de memória e compressão
+├── optimization/   Redução de tokens e custos
+├── teaching/       Modo professor e aprendizado adaptativo
+├── architecture/   Padrões arquiteturais e design
+├── coding/         Skills de engenharia de software
+├── security/       OWASP, autenticação, auditoria
+├── testing/        Testes unitários, integração, E2E
+├── devops/         Docker, CI/CD, infraestrutura
+├── database/       SQL, NoSQL, otimização
+├── frontend/       Skills de frontend
+├── backend/        Skills de backend
+├── package.json    Configuração NPM para publicação CLI
+├── README.md       Documentação principal
+├── SYSTEM.md       Fundação do sistema
+└── RULES.md        Regras operacionais
+```
+
+---
+
+## Agentes e Skills
+
+O framework possui **80+ skills** e **10+ agentes especializados** encadeados dinamicamente via `core/skill-resolver.json`.
+
+```bash
+# Exemplo: Compilar o System Prompt para o Arquiteto de Software
+nexus compile architect prompt_arquiteto.md
+```
+
+---
+
+## Publicando no NPM
+
+```bash
+# Login no NPM
+npm login
+
+# Testar empacotamento em modo dry-run
+npm pack --dry-run
+
+# Publicar pacote publicamente
+npm publish --access public
+```
+
+---
+
+## Licença
+
+MIT — Use, modifique, distribua. Apenas mantenha os créditos.

package/ROADMAP.md

@@ -0,0 +1,131 @@
+# Roadmap
+
+> Vision and planned evolution for the Nexus AI framework.
+
+---
+
+## Legend
+
+| Status | Meaning |
+|--------|---------|
+| ✅ | Done |
+| 🔧 | In progress |
+| 📋 | Planned |
+| 💡 | Future idea |
+
+---
+
+## Phase 1 — Foundation (v1.0.0) ✅
+
+- [x] SYSTEM.md — Core identity and architecture
+- [x] RULES.md — Operating rules and conventions
+- [x] README.md — Entry point and quick start
+- [x] Decision Engine — Task classification and routing
+- [x] Context Engine — Minimal context window builder
+- [x] Token Manager — Budget enforcement and monitoring
+- [x] Compression Engine — 4-level text compression
+- [x] Reflection Engine — Post-task self-review
+- [x] Evolution Engine — Self-improvement from patterns
+- [x] Quality Gates — 5-gate output validation
+- [x] Planning Engine — Task decomposition and effort estimation
+- [x] Memory Manager — 3-tier memory with knowledge graph
+- [x] Professor Mode — Adaptive teaching skill
+- [x] Security Engineer — OWASP + secrets detection
+- [x] Software Architect — Architecture decision skill
+
+---
+
+## Phase 2 — Coding & Database & DevOps (v1.0.0) ✅
+
+### Backend
+- [x] Backend Engineer (Laravel, Node, Python, C#)
+- [x] API Designer — REST, GraphQL, versioning
+- [x] Microservices Expert — patterns, communication, deployment
+
+### Frontend
+- [x] Frontend Engineer (React, Vue, Tailwind)
+- [x] UX Reviewer — usability heuristics
+- [x] Accessibility Reviewer — WCAG compliance
+
+### Database
+- [x] Database Engineer — schema design, migrations
+- [x] SQL Optimizer — query analysis, indexing
+- [x] PostgreSQL Specialist — advanced features
+
+### DevOps
+- [x] DevOps Engineer — infrastructure, deployment
+- [x] Docker Expert — containers, compose, multi-stage
+- [x] CI/CD Specialist — pipelines, automation
+- [x] Cloud Architect (skills/cloud-architect)
+
+---
+
+## Phase 3 — Quality & Testing (v1.0.0) ✅
+
+- [x] Unit Test Engineer — TDD, coverage, mocking
+- [x] Integration Test Engineer — API, database, external services
+- [x] E2E Test Engineer — Cypress, Playwright
+- [x] Clean Code Validator — naming, functions, comments
+- [x] SOLID Validator — LSP, ISP, DIP checks
+- [x] Complexity Analyzer — cyclomatic, cognitive complexity
+- [x] Refactoring Specialist — patterns, migration, risk
+- [x] Technical Debt Analyzer — debt estimation, prioritization
+
+---
+
+## Phase 4 — Advanced Skills (v1.0.0) ✅
+
+- [x] Senior Code Reviewer — PR review with depth
+- [x] Bug Hunter — systematic debugging protocol
+- [x] Root Cause Analyzer — 5 Whys, fishbone
+- [x] Performance Optimizer — profiling, caching, N+1
+- [x] Dependency Analyzer — version conflicts, security, updates
+- [x] Risk Analyzer — probability × impact matrix
+- [x] Breaking Change Detector — semver validation
+- [x] Release Planner — changelog, version bump, tags
+
+---
+
+## Phase 5 — Intelligence (v1.0.0) ✅
+
+- [ ] Knowledge Graph Auto-Builder — extract entities from conversations
+- [x] Hallucination Detector — confidence scoring for statements
+- [x] Confidence Estimator — know when to say "I'm not sure"
+- [x] Alternative Solution Generator — always offer 2+ approaches
+- [x] Trade-off Analyzer — structured comparison with criteria
+- [x] Self-Correction Engine — detect error in own output, fix, log
+- [x] Conversation Summarizer — compress entire session to 200 tokens
+- [x] Context Recovery — resume after session break
+
+---
+
+## Phase 6 — Autonomy & Consolidation (v2.0.0) ✅
+
+- [x] Project Manager — track milestones, deadlines, progress
+- [x] Task Planner — break epics into sprints
+- [x] Requirement Analyzer — extract requirements from natural language
+- [ ] Stakeholder Identifier — who cares about this decision?
+- [x] Cost Optimizer — cloud cost estimation and reduction
+- [x] Documentation Writer — auto-document codebase
+- [ ] UML Diagram Generator — PlantUML from architecture (exists as skill)
+- [x] README Generator — project README from structure
+
+---
+
+## Phase 7 — Ecosystem (v2.0.0 / v3.0.0) 💡
+
+- [ ] Plugin System — third-party skills
+- [ ] Skill Marketplace — share and install skills
+- [ ] Nexus CLI — command-line tool to manage skills
+- [ ] Nexus API — REST API for framework interrogation
+- [ ] Multi-Agent Coordination — multiple Nexus agents collaborating
+- [ ] Web UI — visual skill editor and monitor
+- [ ] Analytics Dashboard — token usage, evolution, cost
+
+---
+
+## Current Focus
+
+**v2.0.0** — Consolidation, plugin system, and CLI tooling to make NexusAI extensible and operable outside of conversation.
+
+> "Rome wasn't built in a day. Neither is a great agent framework."