memexai

TypeScript SDK for MemexAI, a persistent memory service for AI agents.

Install

npm install @sooraj33/memexai

Basic Usage

import { MemexAI } from "@sooraj33/memexai"

const memex = new MemexAI({
  url: "http://localhost:8080",
  apiKey: "dev-agent-key",
})

const memory = memex.forUser({
  userId: "demo_user",
  actor: "assistant",
})

await memory.writeFile({
  path: "user/profile.md",
  content: "# Profile\n\n- Prefers quiet projects near good schools",
  reason: "Captured stable user preference",
})

const profile = await memory.readFile({ path: "user/profile.md" })
console.log(profile.content)

Vercel AI SDK Adapter

import { generateText, stepCountIs } from "ai"
import { createVercelAITools } from "@sooraj33/memexai/adapters/vercel-ai"

const promptBlock = await memory.getPromptBlock()

const result = await generateText({
  model,
  system: [
    "You are a helpful agent with durable memory.",
    "Use memory tools when the user asks you to remember, retrieve, or update stable preferences.",
    "",
    promptBlock,
  ].join("\n"),
  prompt: "Remember that I prefer quiet projects near good schools.",
  tools: createVercelAITools(memory),
  stopWhen: stepCountIs(5),
})

console.log(result.text)

Tool Adapters

The SDK exports adapters for:

• @sooraj33/memexai/adapters/vercel-ai
• @sooraj33/memexai/adapters/openai
• @sooraj33/memexai/adapters/langchain

Links

• Repository: https://github.com/soorajshankar/memexai
• Admin dashboard and local service instructions are in the root README.