What are LLM application patterns?

LLM application patterns are reusable, production-tested architectural templates for building complex AI features such as RAG pipelines, autonomous agents, and prompt management systems.

Which agent architectures are supported?

The skill covers several popular agent patterns including ReAct (Reasoning + Acting), Function Calling, Plan-and-Execute, and Multi-Agent Collaboration.

How does this skill help with RAG implementation?

It provides code patterns for document ingestion, various chunking strategies, vector database selection, and advanced retrieval techniques like hybrid search and reciprocal rank fusion.

Can this skill help with prompt management?

Yes, it includes patterns for creating prompt templates with variables, prompt versioning, and setting up registries for A/B testing different prompts.

LLM Application Patterns

Name: LLM Application Patterns
Author: claudiodearaujo

byclaudiodearaujo

•

データサイエンスとML

Implement production-ready AI architectures using proven patterns for RAG, autonomous agents, and prompt engineering.

LLM Application Patterns provides a comprehensive architectural blueprint for developers building enterprise-grade AI applications. It bridges the gap between basic prompting and sophisticated systems by offering structured guidance on implementing Retrieval-Augmented Generation (RAG) pipelines, selecting optimal agent architectures like ReAct or Plan-and-Execute, and managing complex prompt lifecycles. Whether you are optimizing document ingestion for vector databases, implementing hybrid search, or coordinating multi-agent workflows, this skill provides the implementation patterns and industry best practices required to ensure reliability, scalability, and observability in LLM-powered software.

主な機能

01Guidance on vector database selection and embedding model optimization.

02Advanced retrieval strategies like multi-query retrieval and contextual compression.

03Prompt management patterns for versioning, templating, and A/B testing.

04Standardized RAG pipeline architectures for document ingestion and hybrid search.

05Multiple agent patterns including ReAct, Function Calling, and Plan-and-Execute.

061 GitHub stars

ユースケース

01Building context-aware knowledge bases using RAG and vector search.

02Developing autonomous agents capable of multi-step task planning and tool execution.

03Setting up robust LLMOps monitoring and prompt versioning for production AI apps.

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add claudiodearaujo/sistema-de-narra-o-de-livro llm-app-patterns

For use in Claude.ai and ChatGPT

Download Skill