How does this skill help manage API costs?

It includes strategies for model tiering, embedding caching, and request batching, alongside checklists to avoid expensive anti-patterns like redundant API calls.

Does it handle document processing for RAG applications?

Yes, it includes a robust document chunking utility that manages sentence-aware splitting and overlap to ensure context is preserved for vector embeddings.

Is function calling (tool use) supported?

Yes, it features a complete agent loop pattern that demonstrates how to define tool schemas, handle tool-use stop reasons, and feed results back to the model.

Can I use this for low-latency, real-time applications?

Absolutely. It provides detailed patterns for implementing streaming responses using the Anthropic SDK, allowing for incremental text delivery to users.

What model selection logic is included in this skill?

The skill provides a task-based selection pattern that suggests using specific Claude models like Haiku for classification, Sonnet for coding, and Opus for complex reasoning to balance cost and performance.

LLM Integration Patterns

Name: LLM Integration Patterns
Author: rohitg00

byrohitg00

•

604

•

API Development

Implements production-grade patterns for LLM API usage, streaming, function calling, RAG pipelines, and cost optimization.

This skill provides a comprehensive toolkit for developers building AI-powered applications, offering standardized implementation patterns for the Anthropic SDK. It covers essential architectural needs including real-time streaming responses, robust function calling (tool use), and end-to-end Retrieval-Augmented Generation (RAG) pipelines with intelligent document chunking. Beyond basic integration, the skill focuses on production readiness by providing strategies for cost optimization, model selection based on task complexity, and common anti-patterns to avoid when scaling LLM-driven features.

Key Features

01Production-ready Anthropic API client patterns with streaming support

02Comprehensive function calling and tool use orchestration logic

03Intelligent document chunking algorithms with configurable overlap

04Cost optimization framework via task-specific model selection

05604 GitHub stars

06Complete RAG pipeline architecture with context injection and citation patterns

Use Cases

01Building responsive AI chatbots with real-time streaming and external tool integration

02Developing RAG-based search systems for internal documentation and knowledge bases

03Architecting multi-model systems that optimize for both performance and API budget

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add rohitg00/awesome-claude-code-toolkit llm-integration

For use in Claude.ai and ChatGPT

Download Skill