Crate llm_memory_graph 

LLM-Memory-Graph: Graph-based context-tracking and prompt-lineage database

This crate provides a high-performance, embeddable graph database specifically designed for tracking LLM conversation context, prompt lineage, and multi-agent coordination.

Features

• Context Persistence: Maintain conversation history across sessions
• Prompt Lineage: Track prompt evolution and template inheritance
• Graph-Native: Efficient relationship queries using graph algorithms
• Embedded Storage: Low-latency, file-based storage using Sled
• Type-Safe: Strongly typed nodes and edges with schema validation

Quick Start

use llm_memory_graph::{MemoryGraph, Config};

let config = Config::default();
let graph = MemoryGraph::open(config)?;

let session = graph.create_session()?;
let prompt_id = graph.add_prompt(
    session.id,
    "Explain quantum computing".to_string(),
    None
)?;

Re-exports

pub use engine::AsyncMemoryGraph;

pub use engine::MemoryGraph;

pub use error::Error;

pub use error::Result;

pub use types::AgentConfig;

pub use types::AgentId;

pub use types::AgentMetrics;

pub use types::AgentNode;

pub use types::AgentStatus;

pub use types::Config;

pub use types::ContextType;

pub use types::ConversationSession;

pub use types::EdgeType;

pub use types::InheritsProperties;

pub use types::InstantiatesProperties;

pub use types::InvokesProperties;

pub use types::NodeId;

pub use types::NodeType;

pub use types::Priority;

pub use types::PromptMetadata;

pub use types::PromptNode;

pub use types::PromptTemplate;

pub use types::ReferencesProperties;

pub use types::ResponseMetadata;

pub use types::ResponseNode;

pub use types::SessionId;

pub use types::TemplateId;

pub use types::TokenUsage;

pub use types::ToolInvocation;

pub use types::TransfersToProperties;

pub use types::VariableSpec;

pub use types::Version;

pub use types::VersionLevel;

Modules

engine

Core engine for the memory graph

error

Error types for LLM-Memory-Graph operations

migration

Migration utilities for transitioning from sync to async APIs

observatory

Observatory integration for real-time event streaming and metrics

plugin

Plugin system for extending LLM-Memory-Graph functionality

query

Query interface for graph traversal and filtering

storage

Storage backend for persisting graph data

types

Core data types for LLM-Memory-Graph

Constants

VERSION

Current version of the LLM-Memory-Graph library