crewAIInc/crewAI-examplesArchived
CrewAI Examples
Este proyecto es una librería de referencia y plantillas para la orquestación de múltiples agentes, sistemas de generación aumentada por recuperación (RAG) y patrones de humano en el bucle (human-in-the-loop). Contiene una colección de patrones de implementación para coordinar múltiples agentes de IA para automatizar flujos de trabajo y tareas empresariales complejas.
El proyecto demuestra cómo desacoplar la lógica operativa de la configuración mediante el uso de archivos de texto estructurados para roles de agentes y especificaciones de tareas. Incluye guías para integrar búsqueda vectorial y procesamiento de documentos para construir sistemas de preguntas y respuestas basados en bases de conocimiento externas, así como patrones para insertar pasos de validación y revisión manual en secuencias de agentes automatizadas.
Los ejemplos cubren el diseño de rutas de ejecución iterativas, procesamiento paralelo y enrutamiento dinámico. Estas implementaciones incorporan gestión de estado para rastrear el progreso a través de secuencias complejas y vincular agentes a APIs externas, bases de datos y sistemas de archivos para la ejecución aumentada por herramientas.
Los materiales de referencia se proporcionan como Jupyter Notebooks.
Features
- Role-Based Agent Orchestration - Coordinates multiple specialized AI agents by assigning distinct roles and goals to automate business workflows.
- Multi-Agent Orchestrators - Provides reference implementations for coordinating specialized AI agents to collaborate on complex business workflows.
- Vector Knowledge Bases - Integrates document processing with vector knowledge bases to provide agents with grounded external information.
- Multi-Agent Orchestration Patterns - Offers implementation patterns for coordinating multiple specialized agents to decompose and execute complex workflows.
- AI Workflow Automation - Automates repetitive professional tasks using iterative execution paths and dynamic AI routing.
- Retrieval-Augmented Generation - Implements the retrieval-to-generation pipeline to ground AI responses using external search indices.
- Human-in-the-Loop Systems - Provides tools for integrating human oversight and validation into AI-driven workflows.
- Human-in-the-Loop Workflows - Implements manual review checkpoints and validation steps within automated agent sequences.
- Tool-Augmented Language Models - Extends language model capabilities by linking agents to external APIs, databases, and file systems.
- RAG Implementations - Implements RAG architectures that retrieve relevant documents from vector stores to augment LLM context.
- Iterative Routing - Provides patterns for designing iterative execution paths and dynamic routing to achieve complex goals.
- Agentic Workflows - Provides frameworks for designing autonomous agent processes using iterative refinement and state management.
- External API Tool Exposures - Demonstrates how to expose external API functions as callable tools for AI agents to execute actions.
- YAML-Defined Agents - Defines custom agents and their toolkits using YAML files to decouple configuration from code.
- External Service Integrations - Provides implementations for connecting AI agents to third-party APIs and external platforms to fetch data.
- Workflow State Managements - Includes state management to track progress and context across iterative loops and parallel processing sequences.
- Agent Task State Persistence - Tracks progress and context across iterative loops to maintain consistency during complex agent sequences.
- Iterative Execution Paths - Implements iterative execution paths with dynamic routing and parallel processing to automate complex tasks.
- RAG Implementation Guides - Provides step-by-step frameworks and examples for building retrieval-augmented generation systems.
- YAML Configuration Files - Uses structured YAML files to separate agent definitions and task specifications from execution logic.
- Agent Configuration Files - Defines agent roles and task specifications in YAML files to decouple operational logic from setup.
- Agent Frameworks - Collection of practical implementation examples for multi-agent systems.
- Agent Framework Examples - Collection of role-based multi-agent workflows for business automation.
