skills.rest
JCETools-Petra
JCETools-PetraCOMMUNITY
Explore all skills by this author →
GitHub

ai-llm-engineering

Design RAG pipelines with chunking, embedding, retrieval, and evaluation metrics.

16|Updated Apr 30, 2026
Search Tags:#embeddings#rag#guardrails#prompt-engineering#vector-database#llm-evaluation#citation-grounding
One-click install
npx skills add https://github.com/JCETools-Petra/JCE-Opencode-Tools --skill ai-llm-engineering-jcetools-petra
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: ai-llm-engineering
Source: https://github.com/JCETools-Petra/JCE-Opencode-Tools/tree/main/config/skills/ai-llm-engineering
Command: npx skills add https://github.com/JCETools-Petra/JCE-Opencode-Tools --skill ai-llm-engineering-jcetools-petra

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This Skill helps you build, debug, and evaluate AI/LLM systems that rely on retrieval-augmented generation (RAG), embeddings, vector databases, and prompt engineering—so answers stay grounded in evidence instead of drifting into hallucinations.

Core Features & Use Cases

  • RAG architecture decision-making: choose between naive, advanced, agentic, graph, and streaming RAG based on accuracy and freshness needs.
  • Vector database selection guidance: compare operational trade-offs across pgvector, Pinecone, Qdrant, and hybrid/multimodal options.
  • Production RAG pipeline patterns: chunking strategies, embedding workflows, retrieval with query expansion, deduplication + reranking, and citation-grounded generation.
  • Prompt engineering playbooks: structured outputs, guardrailed prompting, few-shot classification, and self-consistency for better reliability.
  • Evaluation and guardrails: RAGAS-style metrics (faithfulness, relevancy, recall, precision) plus guardrails for injection, PII, and hallucination checks.

Quick Start

Ask the AI to design a RAG pipeline for your knowledge base with chunking, embedding, retrieval (including reranking), and an evaluation plan for faithfulness and context recall.

Frequently Asked Questions about ai-llm-engineering

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I design a RAG pipeline that prevents LLM hallucinations?▼

Build a RAG pipeline with semantic chunking, embedding generation, and citation-grounded generation to ensure LLM answers stay grounded in retrieved evidence instead of hallucinating.

What are the best chunking and embedding strategies for a vector database?▼

Use semantic chunking to split documents and generate embeddings, then apply query expansion with hybrid retrieval and reranking to optimize vector database search accuracy and context recall.

How do I evaluate RAG systems for faithfulness and answer relevancy?▼

Evaluate RAG systems using RAGAS-style metrics including faithfulness, answer relevancy, context recall, and context precision to measure how accurately generated answers match retrieved context.

How do I choose between pgvector, Pinecone, and Qdrant for RAG architecture?▼

Compare vector database options by evaluating operational trade-offs across pgvector, Pinecone, and Qdrant against your specific accuracy, freshness, and multimodal scaling requirements.

What guardrails do I need for prompt engineering in production LLM applications?▼

Implement retrieval-time guardrails for injection detection, PII filtering, and hallucination checks, alongside structured outputs and few-shot classification to ensure production LLM reliability.

Related Skills

Explore More Software Engineering Skills→
PaldomPaldom
1

unstructured-pdf-generation

Generate synthetic PDF documents with JSON evaluation files for RAG systems.

Community
Intermediate
JeffallanJeffallan
10.9k

prompt-engineer

Design and optimize LLM prompts with structured outputs and evaluation frameworks.

Community
Advanced
ryasrkryasrk

prompt-engineering

Design system prompts, RAG pipelines, and tool-calling workflows for LLM providers.

Community
Advanced
AbdullahMalik17AbdullahMalik17
9

memory-systems

Design layered memory architectures for agents using vector stores and temporal knowledge graphs.

Community
Advanced