agentic-workflows

To design an AI agentic workflow, implement established patterns like ReAct, Plan-and-Execute, and tool use to structure an agent's perception, reasoning, and action loops for autonomous task completion. This skill provides the conceptual frameworks and Python snippets needed for implementation.

The ReAct pattern in multi-agent orchestration is a design framework where AI agents cycle through reasoning and acting steps, allowing them to perceive environments, use tools, and learn from outcomes to solve complex tasks autonomously.

Build a customer support AI agent by implementing a reasoning loop that understands user queries, executes tool use to search the knowledge base, and formulates responses based on retrieved findings. This skill provides the structural patterns for that action loop.

Yes, implementing planning agents requires an understanding of agent perception, reasoning, action, and learning loops. This skill provides advanced patterns and guidance, assuming you have foundational knowledge of autonomous AI agent systems.

The best way to structure memory systems for autonomous AI agents is using established architectural patterns that support the agent's reasoning and learning loops. This skill guides the design of memory components within broader multi-agent architectures.

Use multi-agent orchestration when tasks require distributed reasoning and specialized tool use across different agent roles. This approach is necessary when a single workflow cannot handle the complexity of autonomous perception, planning, and action execution.