Crate traitclaw_rag 

RAG (Retrieval-Augmented Generation) pipeline for the TraitClaw AI agent framework.

Provides a Retriever trait, grounding strategies, and a built-in KeywordRetriever with BM25-style scoring for text search.

Quick Start

use traitclaw_rag::{Document, KeywordRetriever, Retriever, GroundingStrategy, PrependStrategy};

let mut retriever = KeywordRetriever::new();
retriever.add(Document::new("doc1", "Rust is a systems programming language"));
retriever.add(Document::new("doc2", "Python is great for AI"));

let docs = retriever.retrieve("Rust systems", 5).await?;
assert!(!docs.is_empty());

let strategy = PrependStrategy;
let context = strategy.ground(&docs);
assert!(context.contains("Rust"));

Re-exports

pub use chunker::Chunker;

pub use chunker::FixedSizeChunker;

pub use chunker::RecursiveChunker;

pub use chunker::SentenceChunker;

pub use embedding::EmbeddingProvider;

pub use embedding::EmbeddingRetriever;

pub use hybrid::CitationStrategy;

pub use hybrid::ContextWindowStrategy;

pub use hybrid::HybridRetriever;

pub use rag_context::RagContextManager;

Modules

chunker

Document chunking strategies for RAG pipelines.

embedding

Embedding-based vector retrieval for RAG pipelines.

hybrid

Hybrid retrieval combining keyword and embedding-based search.

rag_context

RagContextManager — automatic document retrieval and injection.

Structs

Document

A document for retrieval.

KeywordRetriever

BM25-style keyword retriever for in-memory text search.

PrependStrategy

Simple grounding strategy that prepends documents as numbered context.

Traits

GroundingStrategy

Strategy for grounding agent context with retrieved documents.

Retriever

Trait for document retrieval.