Crate rlm_rs 

RLM-RS

Recursive Language Model REPL for Claude Code.

RLM-RS is a CLI tool for handling large context files via chunking and recursive sub-LLM calls. It allows LLMs to process prompts far exceeding their context windows by decomposing content into manageable chunks.

Features

• Chunking: Multiple strategies (fixed, semantic, parallel) for splitting content
• SQLite Storage: Persistent state with transaction support
• Memory Mapping: Efficient handling of large files
• Unicode Aware: Proper grapheme cluster handling

Re-exports

pub use error::Error;

pub use error::Result;

pub use core::Buffer;

pub use core::BufferMetadata;

pub use core::Chunk;

pub use core::ChunkMetadata;

pub use core::Context;

pub use core::ContextValue;

pub use storage::DEFAULT_DB_PATH;

pub use storage::SqliteStorage;

pub use storage::Storage;

pub use chunking::Chunker;

pub use chunking::FixedChunker;

pub use chunking::SemanticChunker;

pub use chunking::available_strategies;

pub use chunking::create_chunker;

pub use cli::Cli;

pub use cli::Commands;

pub use cli::OutputFormat;

pub use embedding::FastEmbedEmbedder;

pub use embedding::DEFAULT_DIMENSIONS;

pub use embedding::Embedder;

pub use embedding::FallbackEmbedder;

pub use embedding::cosine_similarity;

pub use embedding::create_embedder;

pub use search::DEFAULT_SIMILARITY_THRESHOLD;

pub use search::DEFAULT_TOP_K;

pub use search::RrfConfig;

pub use search::SearchConfig;

pub use search::SearchResult;

pub use search::buffer_fully_embedded;

pub use search::embed_buffer_chunks;

pub use search::hybrid_search;

pub use search::reciprocal_rank_fusion;

pub use search::search_bm25;

pub use search::search_semantic;

pub use search::weighted_rrf;

Modules

chunking

Chunking strategies for RLM-RS.

cli

CLI layer for RLM-RS.

core

Core domain models for RLM-RS.

embedding

Embedding generation for semantic search.

error

Error types for RLM-RS operations.

io

I/O utilities for RLM-RS.

search

Hybrid search with semantic and lexical retrieval.

storage

Storage layer for RLM-RS.