Crate agentmem 

Agent Memory RS

A secure, local-first memory layer for AI agents.

This crate is designed to provide a predictable, typed, auditable API for:

• project-scoped local memory
• namespaced agent state
• durable on-disk persistence
• strict validation of keys and paths
• future-safe extension toward multi-agent workflows

Module layout

• error contains crate-wide error types
• types contains domain-specific typed wrappers
• config contains configuration loading and validation ! - core contains foundational validation and namespace logic
• store contains the durable storage engine
• cli contains CLI-facing logic used by binaries

Stability note

This crate is in early development. The goal is to make the behavior predictable before making the public API large.

Re-exports

pub use crate::error::AgentMemoryError;

pub use crate::error::Result;

pub use crate::store::Store;

pub use crate::types::Key;

pub use crate::types::Namespace;

pub use crate::types::ProjectName;

pub use crate::types::StorePath;

pub use crate::types::Value;

Modules

cli

Command-line interface support.

config

core

Core domain primitives and validation logic.

error

index

Local code indexing and retrieval primitives.

store

Durable store layer.

types