sumudunissanka/ai.memory.redis

Overview

This module provides a Redis-backed short-term memory store to use with AI messages (e.g., with AI agents, model providers, etc.).

Key Features

• Redis-backed persistent storage for short-term AI message memory
• Separate storage for system messages (SET/GET) and interactive messages (RPUSH/LRANGE)
• Configurable maximum messages per key with automatic enforcement
• Built-in in-memory caching for improved read performance
• Configurable key prefix for multi-tenant or multi-application isolation
• Support for both connection configuration and existing redis:Client reuse

Prerequisites

• A running Redis server (local or cloud-hosted)

Quickstart

Follow the steps below to use this store in your Ballerina application:

1. Import the ballerinax/ai.memory.redis module.

import ballerinax/ai.memory.redis;

Optionally, import the ballerina/ai and/or ballerinax/redis module(s).

import ballerina/ai;
import ballerinax/redis;

2. Create the short-term memory store by passing either the connection configuration or a redis:Client.

   i. Using the connection configuration

   import ballerina/ai;
   import ballerinax/ai.memory.redis;
   
   configurable string host = ?;
   configurable int port = ?;
   
   ai:ShortTermMemoryStore store = check new redis:ShortTermMemoryStore({
       host, port
   });

   ii. Using an existing redis:Client

   import ballerina/ai;
   import ballerinax/redis;
   import ballerinax/ai.memory.redis as redisStore;
   
   configurable string host = ?;
   configurable int port = ?;
   
   redis:Client redisClient = check new redis:Client(connection = {host, port});
   ai:ShortTermMemoryStore store = check new redisStore:ShortTermMemoryStore(redisClient);

   Optionally, specify the maximum number of messages per key (maxMessagesPerKey - defaults to 20), the in-memory cache configuration (cacheConfig), and a custom key prefix (keyPrefix - defaults to "chat_memory").

   ai:ShortTermMemoryStore store = check new redis:ShortTermMemoryStore({
       host, port
   }, 10, {capacity: 10}, "my_app_memory");