RAGiment
RAGiment
Component registry

RAG Pipelines

Production-ready RAG pipeline archetypes. Each ships with runnable code, config, vector DB setup, and an eval suite.

Mainstream

Mainstream
v1.0.0
LdSpDense SearchBM25 SearchMRerankLLM

Standard RAG

rag-standard

Hybrid vector + BM25 retrieval pipeline. The production baseline.

Latency
Low latency
Database
Vector DB Required
Cost$$ (Medium)
Complexity
Beginner friendly

Best suited for:

  • corpus_size < 10M vectors
  • update_frequency == frequent
langchainllamaindexraw pythontypescript
View component
Mainstream
v1.0.0
LdEntities (NER)RelationshipsGraphCypherTraverseLLM

GraphRAG

rag-graph

Knowledge graph-based retrieval for multi-hop reasoning.

Latency
Medium latency
Database
Neo4j Graph DB
Cost$$$ (High)
Complexity
Advanced

Best suited for:

  • multi-hop reasoning required
  • queries involve entity relationships
langchainllamaindexraw python
View component
Mainstream
v1.0.0
UserReActVector ToolKeyword ToolObserveGroundedLLM

Agentic RAG

rag-agentic

LLM decides when and what to retrieve. Self-directed search.

Latency
Higher latency
Database
Vector DB Required
Cost$$$ (High)
Complexity
Intermediate

Best suited for:

  • query complexity == high
  • multi-step research workflows
langchainllamaindexraw python
View component

Emerging

Emerging
v1.0.0
FileTokensBM25 pklLexicalRankLLM

PageIndex RAG

rag-vectorless

No vector DB. LLM navigates a page tree to find answers.

Latency
High latency
Database
No Database
Cost$$$ (High)
Complexity
Beginner friendly

Best suited for:

  • corpus_size < 500 documents
  • zero infrastructure preference
llamaindexraw python
View component
Emerging
v1.0.0
DocClustersWiki PageCosineSynthesizeLLM

LLM Wiki

rag-wiki

Karpathy-style compounding knowledge base. Gets smarter over time.

Latency
Medium latency
Database
Vector DB Required
Cost$$$ (High)
Complexity
Advanced

Best suited for:

  • corpus_size < 1000 documents
  • corpus is static or slow-changing
raw pythonlangchain
View component
Emerging
v1.0.0
PDFText extractImage extractText StoreImage StoreMLLM

Multi-modal RAG

rag-multimodal

Text + image + PDF vision. Understands documents humans see.

Latency
Higher latency
Database
Vector DB Required
Cost$$$ (High)
Complexity
Advanced

Best suited for:

  • corpus contains PDFs with images/charts/tables
  • queries reference visual content
llamaindexraw pythonlangchain
View component