agent-development

Implement an AI agent by designing an agent loop that processes reasoning steps and integrates external tools. This Skill provides Python code examples for architectures like ReAct and Plan-and-Execute to interact with tools effectively.

ReAct architectures interleave reasoning and acting step-by-step, while Plan-and-Execute separates initial planning from subsequent execution. This Skill covers both patterns, providing code examples to implement these LLM-powered agent workflows.

Implement memory systems for AI agents by integrating conversation history, summary generation, and entity tracking modules. This Skill provides design patterns to maintain context and manage memory for LLM-powered agents effectively.

Yes, you can coordinate multiple LLM agents using orchestrator, debate, or hierarchical patterns. This Skill provides architectural guidance and code examples for designing multi-agent systems that distribute tasks and coordinate actions.

Production considerations for deploying LLM agents include implementing rate limiting and comprehensive logging within the agent loop. This Skill addresses these operational requirements alongside tool definition principles for stable deployments.

Prior prompt engineering experience is recommended to effectively design multi-agent systems. This Skill provides advanced patterns for agent architecture, memory management, and tool integration, requiring foundational LLM interaction knowledge.