nirvana context enricher
Squad de pesquisa paralela que gera contexto enriquecido ao estado nirvana sobre qualquer tópico. Orquestra 3 pesquisadores em paralelo (deep research, skills.sh + bibliotecas GitHub, papers acadêmicos) e 1 sintetizador, com gates mínimos de busca, para produzir um relatório de contexto técnico acionável.
🧘 Nirvana Context Enricher (NCE)
Parallel research squad that generates nirvana-state enriched context on any topic. Evolution of
with Agent Teams: deep research + skills.sh + GitHub libraries + academic papers./atualise
📑 Table of Contents
- Overview
- Comparison: /atualise vs NCE
- Tech Stack
- Prerequisites
- Getting Started
- Architecture
- Agents
- Workflow
- Output
- Troubleshooting
- Contributing
- License
✨ Overview
The Nirvana Context Enricher is a Squad Protocol v5 squad that orchestrates 5 specialized agents to generate comprehensive context reports on any topic. Unlike traditional sequential research, NCE runs 3 researchers in parallel — covering deep research, tools ecosystem, and academic literature — then synthesizes everything into a unified 16-section report.
Who is it for? Developers, architects, and teams who need deep, actionable technical context before making decisions.
Why does it exist? Manual research is slow, shallow, and misses important sources. NCE guarantees minimum coverage of 24 web searches and 11 content extractions, distributed among dedicated researchers1.
📊 Comparison: /atualise vs NCE
| Aspect | /atualise | NCE |
|---|---|---|
| Execution | Sequential (1 agent) | Parallel (4 agents) |
| Rounds | 6 sequential | 3 parallel researchers + synthesis |
| Skills | Does not search | Searches skills.sh |
| GitHub Libraries | No focus | Trending + growth |
| Papers | Mixed with other data | Dedicated researcher |
| Output | Informative report | Actionable enriched context |
| Minimum searches | 15 | 24 |
| Minimum WebFetch | — | 11 |
/atualise
🔧 Tech Stack
| Technology | Purpose |
|---|---|
| Claude Code Agent Teams | Parallel agent orchestration |
| WebSearch | Web search across multiple sources |
| WebFetch | Content extraction and page analysis |
| TeamCreate / TeamDelete | Team lifecycle management |
| TaskCreate / TaskUpdate | Task management and tracking |
| SendMessage | Inter-agent communication |
Agent (run_in_background) | Parallel researcher spawning |
📋 Prerequisites
- Claude Code — with Agent Teams support
- squads-sh — orchestration framework
- Internet access — required for WebSearch and WebFetch
- NCE squad installed — present at
squads/nirvana-context-enricher/
Verify the squad is available:
ls squads/nirvana-context-enricher/squad.yaml⚡ Getting Started
1. Via Skill (recommended)
/nirvana-context-enricher "React Server Components" 20262. Via Squad Command
/SQUADS:nce:nce-orchestrator *enrich "Kubernetes operators" 20263. Arguments
| Argument | Type | Required | Default | Description |
|---|---|---|---|---|
topic | string | ✅ | — | Subject to research |
year | number | ❌ | 2026 | Target year to filter results |
Setup Checklist
- Claude Code open with the squads runtime active
- Topic clearly defined
- Stable internet connection
- Wait for all 3 researchers to complete (~2-5 min)
🏗️ Architecture
Component Diagram
Execution Flow
:file_folder: Expand directory tree
squads/nirvana-context-enricher/
├── squad.yaml # Squad definition
├── README.md # Documentation (PT-BR)
├── agents/
│ ├── nce-orchestrator.md # Nirva — FlowMaster
│ ├── nce-deep-researcher.md # Sage — Seeker
│ ├── nce-skills-scout.md # Scout — Explorer
│ ├── nce-papers-researcher.md # Scholar — Seeker
│ └── nce-synthesizer.md # Lotus — Alchemist
├── tasks/
│ ├── nce-orchestrator-parse-request.md # Parse topic and queries
│ ├── nce-orchestrator-dispatch-research.md # Spawn researchers
│ ├── nce-deep-researcher-execute-rounds.md # 6 research rounds
│ ├── nce-skills-scout-search-ecosystem.md # Skills + libraries
│ ├── nce-papers-researcher-find-papers.md # Academic papers
│ ├── nce-synthesizer-generate-report.md # Final report
│ └── nce-orchestrator-cleanup.md # Shutdown and cleanup
├── workflows/
│ └── nirvana-enrichment.yaml # 4-phase workflow
├── checklists/
│ └── research-quality.md # Quality checklist
├── templates/
│ └── nirvana-report.md # Report template
└── config/
├── coding-standards.md # Coding standards
├── tech-stack.md # Technologies used
└── source-tree.md # Directory structure👥 Agents
| ID | Persona | Archetype | Role | Responsibility |
|---|---|---|---|---|
nce-orchestrator | Nirva | FlowMaster | Orchestrator | Receives topic, analyzes, coordinates 3 parallel researchers + 1 synthesizer |
nce-deep-researcher | Sage | Seeker | Deep Researcher | 6 exhaustive rounds: fundamentals, structure, practice, advanced, current events, problems |
nce-skills-scout | Scout | Explorer | Skills Explorer | skills.sh + GitHub trending libraries + awesome-lists |
nce-papers-researcher | Scholar | Seeker | Academic Researcher | Papers from arxiv/scholar + best approaches and patterns |
nce-synthesizer | Lotus | Alchemist | Nirvana Synthesizer | Combines data from all 3 researchers into the final 16-section report |
Minimum Searches per Agent
| Agent | WebSearch | WebFetch | Total |
|---|---|---|---|
| Sage | 10 | 5 | 15 |
| Scout | 8 | 3 | 11 |
| Scholar | 6 | 3 | 9 |
| Total | 24 | 11 | 35 |
minimum_viable gate (2/3 researchers) is triggered.
🔄 Workflow
The nirvana-enrichment workflow has 4 sequential phases with the research phase running in parallel:
Phases
| # | Phase | Type | Agent(s) | Dependency |
|---|---|---|---|---|
| 1 | parse | Sequential | Nirva | — |
| 2 | research | Parallel | Sage, Scout, Scholar | parse |
| 3 | synthesize | Sequential | Lotus | research |
| 4 | cleanup | Sequential | Nirva | synthesize |
Error Handling
| Scenario | Behavior |
|---|---|
researcher_failure | Continue with remaining researchers |
minimum_viable | Requires success from at least 2/3 researchers |
timeout | Use partial data collected so far |
Quality Gates
gates:
min_searches: 20 # Minimum total WebSearch
min_webfetch: 11 # Minimum total WebFetch
min_researchers_success: 2 # Minimum successful researchers📄 Output
The report is automatically saved to:
.claude/context-enrichment/{slug}-{YYYY-MM-DD}.md16 Sections of the Nirvana Report
:scroll: View all 16 sections
| # | Section | Primary Source |
|---|---|---|
| 1 | Request Analysis | Nirva |
| 2 | What It Is | Sage |
| 3 | Architecture/Structure | Sage |
| 4 | Core Features | Sage |
| 5 | How to Use | Sage |
| 6 | Configuration | Sage |
| 7 | Best Practices | Sage |
| 8 | Year's Updates | Sage |
| 9 | Recommended Skills (skills.sh) | Scout |
| 10 | Modern Ecosystem Libraries | Scout |
| 11 | Reference Papers & Articles | Scholar |
| 12 | Best Approaches | Scholar |
| 13 | Common Problems and Solutions | Sage + Scholar |
| 14 | Enriched Context (Nirvana) | Lotus |
| 15 | Resources | All |
| 16 | Knowledge Gaps | Lotus |
Quality Rules (Checklist)
| Category | Type | Criteria |
|---|---|---|
| Quantity | Blocking | Minimum searches/webfetch per agent |
| Quality | Blocking | Official sources, year filter, 2/3 researchers, registered URLs |
| Completeness | Advisory | 16 sections filled, original Enriched Context, gaps documented |
| Output | Blocking | File saved, cleanup executed, result reported |
🆘 Troubleshooting
| Problem | Probable Cause | Solution |
|---|---|---|
| Incomplete report | Researcher failed | Check if at least 2/3 succeeded (minimum_viable gate) |
| Research timeout | Slow connection or topic too broad | Refine the topic to be more specific |
| Empty sections | WebFetch blocked by site | Check URLs in log; Lotus marks as gap |
| TeamCreate error | Agent limit reached | Wait for other squads to finish or cancel with Ctrl+C |
| File not generated | Cleanup ran before synthesis | Verify that min_researchers_success >= 2 was met |
:mag: Detailed diagnostics
Check squad status
ls -la .claude/context-enrichment/Verify the squads runtime recognizes the squad
ls squads/nirvana-context-enricher/squad.yamlForce manual cleanup
# If automatic cleanup fails, check for residual agents
# and cancel manually via Claude CodeCheck partial output
# Partial reports are saved with -partial suffix
ls .claude/context-enrichment/*-partial*🤝 Contributing
Contributions are welcome! Follow the steps below:
- Fork the project
- Create your branch ()
git checkout -b feature/nce-improvement
- Commit your changes (
git commit -m 'feat(nce): improvement description') - Push to the branch ()
git push origin feature/nce-improvement
- Open a Pull Request
Commit Patterns
| Type | Description |
|---|---|
feat(nce): | New NCE feature |
fix(nce): | Bug fix |
docs(nce): | Documentation update |
refactor(nce): | Code refactoring |
test(nce): | Test addition or fix |
chore(nce): | Maintenance tasks |
See the Contributing Guide for more details2.
:newspaper: Changelog
v1.0.0 (2026-03-06)
+ Added: NCE squad with 5 specialized agents
+ Added: nirvana-enrichment workflow with 4 phases
+ Added: Parallel research with 3 simultaneous researchers
+ Added: Nirvana report with 16 sections
+ Added: Quality gates with search minimums
+ Added: Error handling with minimum_viable (2/3)
! Changed: Evolution from /atualise to multi-agent architecture📃 License
This project is licensed under the MIT license — see the LICENSE file for details.
Made with ❤️ by Luiz Gustavo Vieira Rodrigues
⭐ If this project helped you, consider giving it a star!
Footnotes
Reviews
0 reviewsNo reviews yet. Be the first to review this squad!
More from Luiz Rodrigues
squad forge
Squad meta de desenvolvimento, otimização e evolução de squads agênticos — cria, valida, otimiza e moderniza qualquer artefato de squad (agentes, tasks, workflows, skills, squads inteiros, templates) com perfeição de formato, e mantém a própria base de conhecimento atualizada via pesquisa web.
ultimate landingpage
Squad fullstack para criação de landing pages de ultra-alta conversão — pipeline end-to-end com 9 agentes: discovery de produto, pesquisa de mercado e experts de copywriting, design system atômico com contraste WCAG AAA e light/dark, geração de imagens por IA, frontend Next.js com SEO e acessibilidade, backend Python FastAPI com admin panel e exportação CSV, integrações WhatsApp e email, e QA multidimensional com score por dimensão
data quality guardian
Squad especialista em qualidade de dados — profiling de datasets, detecção de anomalias, validação de schemas, geração de relatórios de qualidade e sugestão de remediações automatizadas para pipelines de dados
incident response squad
Squad especialista em resposta a incidentes para DevOps/SRE — análise de logs multi-source, correlação de causa raiz, execução de runbooks de remediação, comunicação de status e geração de post-mortems blameless
brandcraft
Squad de produção de entregáveis visuais brand-consistent: extrai design systems de URLs (Refero + live extraction), gera PDFs, PPTX, posts sociais, carrosséis e vídeos programáticos por dois caminhos (Veo 3.1 + Remotion para footage de IA; HyperFrames HTML-first determinístico para motion-design, data-viz e explainer), faz auditoria competitiva de marca, aplica um quality gate awwwards de 24 itens e um arsenal de 112 bibliotecas de UI nos HTMLs/PDFs/animações, e valida acessibilidade WCAG 2.2 AA. Pipeline com 16 agentes especializados e 16 workflows.
nirvana squad creator
Gera squads v5-compliant a partir de linguagem natural — pipeline de 11 fases começando com intent archaeology + pesquisa web (2026), depois análise grounded, geração, otimização, validação, README multi-idioma, deploy e publicação no squads.sh
notebooklm automation
Automação programática completa do Google NotebookLM — da criação de notebooks e curadoria de fontes à pesquisa por Q&A progressivo e geração de 10 formatos de conteúdo (podcast, vídeo, slides, quiz, flashcards, infográfico, mapa mental, tabela de dados, relatório), com download de artefatos orquestrado por pipeline inteligente. Suporta 50+ idiomas e error handling cross-cutting (retry, re-auth, fallback de cota).
nirvana readme architect
Squad de geração de README.md perfeito e impecável, combinando análise profunda de codebase, seleção de template por tipo de projeto, todas as features do GitHub Flavored Markdown, validação com checklist de 25+ pontos e polimento final com badges e TOC.
soc alert triage
Squad especialista em triagem de alertas SOC para cybersecurity — classificação automatizada, filtragem de falsos positivos, priorização de ameaças, enriquecimento com threat intel e geração de briefs para analistas
adaptive tutor k12
Squad especialista em tutoria adaptativa K-12 — avaliação diagnóstica, mapeamento curricular personalizado, sessões de tutoria com dificuldade adaptativa, rastreamento de progresso e relatórios para pais e educadores
resume screener squad
Squad especialista em triagem de currículos para recrutamento — parsing automatizado de CVs, matching de skills com requisitos da vaga, auditoria de vieses, ranking de candidatos e geração de resumos executivos para hiring managers
automated code review squad
Squad especialista em code review automatizado — revisão de segurança, análise lógica, verificação de padrões arquiteturais, enforcement de coding standards e geração de review summaries priorizados para PRs e commits