rag-implementation
Build Retrieval-Augmented Generation (RAG) systems for LLM applications with vector databases and semantic search. Use when implementing knowledge-grounded AI, building document Q&A systems, or integrating LLMs with external knowledge bases.
81
Does it follow best practices?
If you maintain this skill, you can automatically optimize it using the tessl CLI to improve its score:
npx tessl skill review --optimize ./path/to/skillValidation for skill structure
Evaluation results
82%
75%Product Knowledge Base Q&A System
LangGraph RAG with hybrid search
Voyage embeddings model
0%
100%
Claude Sonnet model
0%
100%
LangGraph StateGraph
0%
100%
Chroma vector store
25%
100%
RecursiveCharacterTextSplitter
0%
100%
Chunk size around 1000
0%
0%
Overlap 10-20%
33%
100%
EnsembleRetriever hybrid
25%
100%
Hybrid search weights
0%
100%
Retriever k=10
0%
0%
Without context: $0.8332 · 6m 51s · 24 turns · 177 in / 13,536 out tokens
With context: $1.0135 · 13m 57s · 31 turns · 521 in / 11,315 out tokens
27%
7%Technical API Documentation Assistant
Parent document retriever with citations
ParentDocumentRetriever
0%
0%
Child chunk size 400
0%
0%
Child overlap 50
0%
0%
Parent chunk size 2000
0%
0%
Parent overlap 200
0%
0%
Citation format in prompt
50%
20%
Citations in output
0%
0%
Source metadata
0%
100%
Date metadata
0%
0%
Cannot answer fallback
100%
100%
Without context: $0.8374 · 7m 48s · 27 turns · 209 in / 12,722 out tokens
With context: $2.0880 · 26s · 1 turns · 10 in / 91 out tokens
50%
28%Compliance Knowledge Base Query Engine
HyDE retrieval with reranking and structured output
HyDE generation step
0%
0%
HyDE used for retrieval
0%
0%
LangGraph StateGraph
0%
100%
CrossEncoder from sentence_transformers
0%
0%
ms-marco rerank model
0%
0%
Structured response: answer
100%
100%
Structured response: confidence
40%
100%
Structured response: sources
100%
100%
Structured response: reasoning
0%
100%
Structured output binding
0%
0%
Without context: $1.0019 · 17m 43s · 33 turns · 241 in / 14,082 out tokens
With context: $2.2291 · 30m 53s · 53 turns · 1,059 in / 22,388 out tokens
- Evaluated
- Agent
- Claude Code
Table of Contents
Is this your skill?
If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.