agent-orchestrator
Meta-skill que orquestra todos os agentes do ecossistema. Scan automatico de skills, match por capacidades, coordenacao de workflows multi-skill e registry management.
What this skill does
# Agent Orchestrator
## Overview
Meta-skill que orquestra todos os agentes do ecossistema. Scan automatico de skills, match por capacidades, coordenacao de workflows multi-skill e registry management.
## When to Use This Skill
- When you need specialized assistance with this domain
## Do Not Use This Skill When
- The task is unrelated to agent orchestrator
- A simpler, more specific tool can handle the request
- The user needs general-purpose assistance without domain expertise
## How It Works
Meta-skill que funciona como camada central de decisao e coordenacao para todo
o ecossistema de skills. Faz varredura automatica, identifica agentes relevantes
e orquestra multiplos skills para tarefas complexas.
## Principio: Zero Intervencao Manual
- **SEMPRE faz varredura** antes de processar qualquer solicitacao
- Novas skills sao **auto-detectadas e incluidas** ao criar SKILL.md em qualquer subpasta
- Skills removidas sao **auto-excluidas** do registry
- Nenhum comando manual e necessario para registrar novas skills
---
## Workflow Obrigatorio (Toda Solicitacao)
Execute estes passos ANTES de processar qualquer request do usuario.
Os scripts usam paths relativos automaticamente - funciona de qualquer diretorio.
## Passo 1: Auto-Discovery (Varredura)
```bash
python agent-orchestrator/scripts/scan_registry.py
```
Ultra-rapido (<100ms) via cache de hashes MD5. So re-processa arquivos alterados.
Retorna JSON com resumo de todos os skills encontrados.
## Passo 2: Match De Skills
```bash
python agent-orchestrator/scripts/match_skills.py "<solicitacao do usuario>"
```
Retorna JSON com skills ranqueadas por relevancia. Interpretar o resultado:
| Resultado | Acao |
|:-----------------------|:--------------------------------------------------------|
| `matched: 0` | Nenhum skill relevante. Operar normalmente sem skills. |
| `matched: 1` | Um skill relevante. Carregar seu SKILL.md e seguir. |
| `matched: 2+` | Multiplos skills. Executar Passo 3 (orquestracao). |
## Passo 3: Orquestracao (Se Matched >= 2)
```bash
python agent-orchestrator/scripts/orchestrate.py --skills skill1,skill2 --query "<solicitacao>"
```
Retorna plano de execucao com padrao, ordem dos steps e data flow entre skills.
## Passo Rapido (Atalho)
Para queries simples, os passos 1+2 podem ser combinados em sequencia:
```bash
python agent-orchestrator/scripts/scan_registry.py && python agent-orchestrator/scripts/match_skills.py "<solicitacao>"
```
---
## Skill Registry
O registry vive em:
```
agent-orchestrator/data/registry.json
```
## Locais De Busca
O scanner procura SKILL.md em:
1. `.claude/skills/*/` (skills registradas no Claude Code)
2. `*/` (skills standalone no top-level)
3. `*/*\` (skills em subpastas, ate profundidade 3)
## Metadata Por Skill
Cada entrada no registry contem:
| Campo | Descricao |
|:---------------|:---------------------------------------------------|
| name | Nome da skill (do frontmatter YAML) |
| description | Descricao completa (triggers inclusos) |
| location | Caminho absoluto do diretorio |
| skill_md | Caminho absoluto do SKILL.md |
| registered | Se esta em .claude/skills/ (true/false) |
| capabilities | Tags de capacidade (auto-extraidas + explicitas) |
| triggers | Keywords de ativacao extraidas da description |
| language | Linguagem principal (python/nodejs/bash/none) |
| status | active / incomplete / missing |
## Comandos Do Registry
```bash
## Scan Rapido (Usa Cache De Hashes)
python agent-orchestrator/scripts/scan_registry.py
## Tabela De Status Detalhada
python agent-orchestrator/scripts/scan_registry.py --status
## Re-Scan Completo (Ignora Cache)
python agent-orchestrator/scripts/scan_registry.py --force
```
---
## Algoritmo De Matching
Para cada solicitacao, o matcher pontua skills usando:
| Criterio | Pontos | Exemplo |
|:-----------------------------|:-------|:--------------------------------------|
| Nome do skill na query | +15 | "use web-scraper" -> web-scraper |
| Keyword trigger exata | +10 | "scrape" -> web-scraper |
| Categoria de capacidade | +5 | data-extraction -> web-scraper |
| Sobreposicao de palavras | +1 | Palavras da query na description |
| Boost de projeto | +20 | Skill atribuida ao projeto ativo |
Threshold minimo: 5 pontos. Skills abaixo disso sao ignoradas.
## Match Com Projeto
```bash
python agent-orchestrator/scripts/match_skills.py --project meu-projeto "query aqui"
```
Skills atribuidas ao projeto recebem +20 de boost automatico.
---
## Padroes De Orquestracao
Quando multiplos skills sao relevantes, o orchestrator classifica o padrao:
## 1. Pipeline Sequencial
Skills formam uma cadeia onde o output de uma alimenta a proxima.
**Quando:** Mix de skills "produtoras" (data-extraction, government-data) e "consumidoras" (messaging, social-media).
**Exemplo:** web-scraper coleta precos -> whatsapp-cloud-api envia alerta
```
user_query -> web-scraper -> whatsapp-cloud-api -> result
```
## 2. Execucao Paralela
Skills trabalham independentemente em aspectos diferentes da solicitacao.
**Quando:** Todas as skills tem o mesmo papel (todas produtoras ou todas consumidoras).
**Exemplo:** instagram publica post + whatsapp envia notificacao (ambos recebem o mesmo conteudo)
```
user_query -> [instagram, whatsapp-cloud-api] -> aggregated_result
```
## 3. Primario + Suporte
Uma skill principal lidera; outras fornecem dados de apoio.
**Quando:** Uma skill tem score muito superior as demais (>= 2x).
**Exemplo:** whatsapp-cloud-api envia mensagem (primario) + web-scraper fornece dados (suporte)
```
user_query -> whatsapp-cloud-api (primary) + web-scraper (support) -> result
```
## Detalhes Em `References/Orchestration-Patterns.Md`
---
## Gerenciamento De Projetos
Atribuir skills a projetos permite boost de relevancia e contexto persistente.
## Arquivo De Projetos
```
agent-orchestrator/data/projects.json
```
## Operacoes
**Criar projeto:**
Adicionar entrada ao projects.json:
```json
{
"name": "nome-do-projeto",
"created_at": "2026-02-25T12:00:00",
"skills": ["web-scraper", "whatsapp-cloud-api"],
"description": "Descricao do projeto"
}
```
**Adicionar skill a projeto:** Atualizar o array `skills` do projeto.
**Remover skill de projeto:** Remover do array `skills`.
**Consultar skills do projeto:** Ler o projects.json e listar skills atribuidas.
---
## Adicionando Novas Skills
Para adicionar uma nova skill ao ecossistema:
1. Criar uma pasta em qualquer lugar sob `skills root:`
2. Criar um `SKILL.md` com frontmatter YAML:
```yaml
---
name: minha-nova-skill
description: "Descricao com keywords de ativacao..."
---
## Documentacao Da Skill
```
3. **Pronto!** O auto-discovery detecta automaticamente na proxima solicitacao.
Opcionalmente, para discovery nativo do Claude Code:
4. Copiar o SKILL.md para `.claude/skills/<nome>/SKILL.md`
## Tags De Capacidade Explicitas (Opcional)
Adicionar ao frontmatter para matching mais preciso:
```yaml
capabilities: [data-extraction, web-automation]
```
---
## Ver Status De Todos Os Skills
```bash
python agent-orchestrator/scripts/scan_registry.py --status
```
## Interpretar Status
| Status | Significado |
|:-----------|:---------------------------------------------------|
| active | SKILL.md com name + description presentes |
| incomplete | SKILL.md existe mas falta name ou description |
| missing | Diretorio existe mas sem SKILL.md |
---
## Skills Atuais Do Ecossistema
| Skill | CapacidRelated in AI Agents
skill-development
IncludedComprehensive meta-skill for creating, managing, validating, auditing, and distributing Claude Code skills and slash commands (unified in v2.1.3+). Provides skill templates, creation workflows, validation patterns, audit checklists, naming conventions, YAML frontmatter guidance, progressive disclosure examples, and best practices lookup. Use when creating new skills, validating existing skills, auditing skill quality, understanding skill architecture, needing skill templates, learning about YAML frontmatter requirements, progressive disclosure patterns, tool restrictions (allowed-tools), skill composition, skill naming conventions, troubleshooting skill activation issues, creating custom slash commands, configuring command frontmatter, using command arguments ($ARGUMENTS, $1, $2), bash execution in commands, file references in commands, command namespacing, plugin commands, MCP slash commands, Skill tool configuration, or deciding between skills vs slash commands. Delegates to docs-management skill for official documentation.
reprompter
IncludedTransform messy prompts into well-structured, effective prompts — single or multi-agent. Use when: "reprompt", "reprompt this", "clean up this prompt", "structure my prompt", rough text needing XML tags and best practices, "reprompter teams", "repromptception", "run with quality", "smart run", "smart agents", multi-agent tasks, audits, parallel work, anything going to agent teams. Don't use when: simple Q&A, pure chat, immediate execution-only tasks. See "Don't Use When" section for details. Outputs: Structured XML/Markdown prompt, quality score (before/after), optional team brief + per-agent sub-prompts, agent team output files. Success criteria: Single mode quality score ≥ 7/10; Repromptception per-agent prompt quality score 8+/10; all required sections present, actionable and specific.
adaptive-compaction
IncludedAdaptive add-on policy and recovery layer that decides WHEN to compact, prune, snapshot, or fork -- replacing fixed-percent auto-compaction across Claude Code, Codex, and MCP-capable hosts. Trigger on auto-compact timing or damage: "when should I compact", "is it safe to compact now or start a fresh session", "auto-compact fires too early/mid-task", "switching to an unrelated task but the window still has space", "context rot", "answers get worse the longer the session runs", "the agent forgot the plan or my decisions after it summarized", "add a layer on top that manages context without changing the agent", raising autoCompactWindow to give the policy room, or installing/tuning a cross-tool compaction policy or PreCompact hook -- even when "compaction" is never said but the problem is context-window pressure or post-summarization memory loss. Do NOT use to summarize a conversation, build RAG, write a summarization prompt (decides WHEN not HOW), or answer max-context-length trivia.
agent-skill-creator
IncludedCreate cross-platform agent skills from workflow descriptions. Activates when users ask to create an agent, automate a repetitive workflow, create a custom skill, or need advanced agent creation. Triggers on phrases like create agent for, automate workflow, create skill for, every day I have to, daily I need to, turn process into agent, need to automate, create a cross-platform skill, validate this skill, export this skill, migrate this skill. Supports single skills, multi-agent suites, transcript processing, template-based creation, interactive configuration, cross-platform export, and spec validation.
llm-wiki
IncludedUse when building or maintaining a persistent personal knowledge base (second brain) in Obsidian where an LLM incrementally ingests sources, updates entity/concept pages, maintains cross-references, and keeps a synthesis current. Triggers include "second brain", "Obsidian wiki", "personal knowledge management", "ingest this paper/article/book", "build a research wiki", "compound knowledge", "Memex", or whenever the user wants knowledge to accumulate across sessions instead of being re-derived by RAG on every query.
skill-master
IncludedAgent Skills authoring, evaluation, and optimization. Create, edit, validate, benchmark, and improve skills following the agentskills.io specification. Use when designing SKILL.md files, structuring skill folders (references, scripts, assets), ingesting external documentation into skills, running trigger evals, benchmarking skill quality, optimizing descriptions, or performing blind A/B comparisons. Keywords: agentskills.io, SKILL.md, skill authoring, eval, benchmark, trigger optimization.