ulmemory schedule add-proactive        # Cada 30 min
ulmemory schedule add-researcher        # Cada hora
ulmemory schedule add-consolidator     # Diario 5am

# Buscar en memoria + web
ulmemory agent run researcher "AI agents" --web

# Especificar fuentes
ulmemory agent run researcher "topic" --sources web,memory,codewiki

# Deep research con expansión de queries
ulmemory agent run researcher "topic" --deep

ulmemory agent skills                    # Ver skills disponibles
ulmemory agent skills researcher         # Ver skills de un agente
ulmemory agent add-skill mi-agente web_search
ulmemory agent edit mi-agente --schedule "0 9 * * *"

# Investigación profunda con web + codewiki
ulmemory agent run auto-researcher "topic:AI,topic:ML" --deep

# Requiere API key de Tavily (gratuita)
export TAVILY_API_KEY="tvly-xxx"

# O en config.yaml:
# research:
#   tavily:
#     api_key: "tvly-xxx"

# Buscar repositorios relacionados
ulmemory agent run researcher "RAG frameworks" --sources codewiki

# Auto-researcher incluye CodeWiki automáticamente
ulmemory agent run auto-researcher "vector databases"

# Combinar todas las fuentes
ulmemory agent run researcher "LLM agents" --sources web,memory,codewiki

# Deep research con expansión automática de queries
ulmemory agent run researcher "RAG systems" --deep

# 1. Iniciar servicios
ulmemory up

# 2. Indexar documentos
ulmemory memory add ./docs/

# 3. Buscar en memoria + web
ulmemory agent run researcher "importante" --web

# 4. Deep research
ulmemory agent run researcher "AI memory systems" --deep --sources web,codewiki

# 5. Analizar salud
ulmemory memory analyze

# 6. Limpiar si es necesario
ulmemory memory consolidate

# 7. Programar mantenimiento diario
ulmemory schedule add consolidator --cron "0 3 * * *" --name "limpieza-diaria"

# 8. Programar research semanal
ulmemory schedule add auto-researcher --cron "0 9 * * 1" --args "topic:AI" --name "research-semanal"

# Limpieza diaria a las 3am
ulmemory schedule add consolidator --cron "0 3 * * *" --name "limpieza-diaria"

# Investigación semanal los lunes a las 9am
ulmemory schedule add auto-researcher --cron "0 9 * * 1" --args "topic:AI,topic:ML" --name "research-semanal"

# Búsqueda cada 6 horas
ulmemory schedule add researcher --cron "0 */6 * * *" --args "updates"

┌───────────── minuto (0-59)
│ ┌───────────── hora (0-23)
│ │ ┌───────────── día del mes (1-31)
│ │ │ ┌───────────── mes (1-12)
│ │ │ │ ┌───────────── día de la semana (0-6, 0=domingo)
│ │ │ │ │
* * * * *

# Ver todas las tareas
ulmemory schedule list

# Ver historial de ejecuciones
ulmemory schedule history 1

# Editar horario
ulmemory schedule edit 1 --cron "0 4 * * *"

# Ejecutar inmediatamente
ulmemory schedule run 1

llm:
  default_provider: "minimax"
  providers:
    minimax:
      api_key: "sk-cp-xxx"
      model: "MiniMax-Text-01"
    google:
      api_key: "AIza-xxx"
      model: "gemini-1.5-flash"
    openai:
      api_key: "sk-xxx"
      model: "gpt-4"

# Research Tools
research:
  tavily:
    api_key: "tvly-xxx"
    enabled: true
  codewiki:
    enabled: true
    cli_path: "~/.claude/skills/codewiki/codewiki"

# Agent Configuration
agents:
  researcher:
    enabled: true
    sources: ["memory", "web", "codewiki"]
  auto_researcher:
    enabled: true
    sources: ["web", "codewiki"]

import asyncio
from core.memory import MemorySystem
from agents.researcher import ResearcherAgent
from agents.tools import registry, WebSearchTool

async def main():
    memory = MemorySystem()

    # Enhanced researcher with web search
    researcher = ResearcherAgent(
        memory,
        enable_web_search=True,
        tavily_api_key="tvly-xxx"
    )

    # Multi-source research
    result = await researcher.research(
        "AI agents",
        sources=["web", "memory", "codewiki"]
    )

    print(f"Web results: {len(result.web_results)}")
    print(f"Memory results: {len(result.memory_results)}")
    print(f"CodeWiki repos: {len(result.codewiki_results)}")

    if result.web_answer:
        print(f"AI Answer: {result.web_answer}")

    # Deep research
    deep_result = await researcher.deep_research(
        "vector databases",
        max_depth=3,
        save_to_memory=True
    )
    print(f"Total sources: {deep_result['total_sources']}")

asyncio.run(main())