Skip to main content

Overview

Walrus Memory exposes SDK surfaces for TypeScript and Python. The SDKs give your agents portable memory that works across apps, sessions, and workflows, fully under your control.

@mysten-incubation/memwal​

Use this first.

  • relayer-backed
  • best path for most teams
  • main methods: remember, recall, analyze, restore, health
import { MemWal } from "@mysten-incubation/memwal";

@mysten-incubation/memwal/manual​

Use this when the client must handle embeddings and local Seal operations.

  • relayer still handles upload relay, registration, search, and restore
import { MemWalManual } from "@mysten-incubation/memwal/manual";

@mysten-incubation/memwal/ai​

Use this when you already use the AI SDK.

import { withMemWal } from "@mysten-incubation/memwal/ai";

memwal​

The Python SDK mirrors the TypeScript MemWal client exactly, same methods, same relayer, same auth flow. Built for the Python AI/ML ecosystem.

  • relayer-backed (same managed relayer endpoints)
  • Ed25519 signing through PyNaCl
  • async-native with a sync convenience wrapper
  • LangChain and OpenAI SDK middleware included
$ pip install memwal
$ pip install memwal[langchain]   # LangChain middleware
$ pip install memwal[openai]      # OpenAI SDK middleware
$ pip install memwal[all]         # everything
from memwal import MemWal

memwal = MemWal.create(
    key="<your-ed25519-private-key>",
    account_id="<your-memwal-account-id>",
    server_url="https://relayer.memory.walrus.xyz",
    namespace="demo",
)

Main methods: remember, recall, analyze, ask, restore, health

Middleware: with_memwal_langchain, with_memwal_openai

Namespace​

All clients support a default namespace. If you omit it, it falls back to "default".

  1. Start with MemWal (TypeScript) or memwal (Python)
  2. Set a namespace explicitly
  3. Validate remember, recall, analyze, and restore
  4. Move to MemWalManual only if you need client-managed embeddings and local Seal work