Describe agent design patterns and workflows for building AI agents.
npx skills add https://github.com/itsmostafa/llm-engineering-skills --skill agents-itsmostafa
Please help me install this Agent Skill. Skill: agents Source: https://github.com/itsmostafa/llm-engineering-skills/tree/main/skills/agents Command: npx skills add https://github.com/itsmostafa/llm-engineering-skills --skill agents-itsmostafa
Patterns and architectures for building AI agents and workflows with LLMs. Use when designing systems that involve tool use, multi-step reasoning, autonomous decision-making, or orchestration of LLM-driven tasks.
Define a task you want the agent to perform.
High-intent search queries and answers about installing and using this skill.
Autonomous AI agents utilize design patterns like ReAct for multi-step reasoning, Orchestrator-Workers for task delegation, and Human-in-the-Loop for controlled execution. These patterns enable LLM-driven processes to use tools and make autonomous decisions.
To implement the ReAct pattern, define a task for the agent to perform, then apply the available tool design architectures. This enables rapid discovery of workflows where the LLM reasons through steps and uses tools to execute actions.
Use an Orchestrator-Workers architecture when designing systems that require orchestration of multiple LLM-driven processes. It allows a central orchestrator to delegate autonomous tasks to worker agents, enabling complex multi-step reasoning.
Add human-in-the-loop oversight by integrating safety considerations and best practices into your agent architectures. This ensures that autonomous decision-making and tool use within LLM workflows remain supervised and controlled.
Safety considerations for LLM-driven autonomous agents involve applying best practices for tool design and workflow orchestration. This ensures that multi-step reasoning and autonomous decision-making processes operate safely within defined boundaries.
This skill applies to LLM orchestration and autonomous agent design patterns, but it focuses on the ReAct reasoning pattern rather than the React front-end framework. It specifically targets tool use and multi-step reasoning workflows.
raphaelchristiConfigure Harness Evolver, LangSmith, and baseline evaluations for Python projects.
DhumitechDesign autonomous AI agent architectures for production systems.
vignesh2027Design and orchestrate end-to-end agentic workflows with LLMs, tools, and human oversight.
slgoodrichPrioritize features using RICE, ICE, MoSCoW, Kano, and weighted scoring frameworks.
