Choose Your Path
Walrus Memory supports several integration modes depending on how much control you need. Pick the one that fits your trust model, tech stack, and control requirements: the Default SDK and managed relayer are the fastest starting point for most teams, the manual client flow and a self-hosted relayer minimize trust in the relayer, and the AI Middleware and MCP Clients match your stack if you already use the AI SDK or an MCP-aware agent.
These paths aren't mutually exclusive. You can combine them - for example, use the Default SDK with the AI Middleware, or start with the Managed Relayer and move to Self-Hosting later. They all share the same backend and data layer.
1. Default SDK
Use @mysten-incubation/memwal when you want the fastest working integration.
- relayer handles embedding, retrieval, and restore
- best starting point for most teams
Go to: SDK Overview
2. Managed relayer
Use a hosted relayer, or deploy your own self-hosted relayer with access to a wallet funded with WAL and SUI.
Following endpoints are provided as public good by Walrus Foundation.
| Network | Relayer URL |
|---|---|
| Production (Mainnet) | https://relayer.memory.walrus.xyz |
| Staging (Testnet) | https://relayer.dev.memwal.ai |
Go to: Managed Relayer
3. Manual client flow
Use @mysten-incubation/memwal/manual when you want full client-side control over encryption and embeddings. Recommended for Web3-native users who want to minimize trust in the relayer - it never sees your plaintext data.
- client handles embeddings and Seal encryption locally
- relayer only sees encrypted payloads and vectors
Go to: SDK Usage
4. AI Middleware
Use @mysten-incubation/memwal/ai when you already use the AI SDK and want recall plus auto-save behavior.
Go to: AI Integration
5. Self-hosted relayer
Use this when you need full control over the trust boundary - your infrastructure, your credentials, no third party sees your data. Self-hosting is a deployment option of the relayer described in path 2, not a different backend: all paths share the same backend and data layer.
Go to: Self-Hosting
6. MCP Clients
Use Walrus Memory's MCP server when you want Cursor, Claude Desktop, Claude Code, Antigravity, or another MCP-aware agent to save and recall memory during tool use.
- connect directly to the hosted relayer with Streamable HTTP at
/api/mcp - or run the local stdio package with
npx -y @mysten-incubation/memwal-mcp
Go to: MCP
If the agent already has local memory, standing context, or session history, copy those facts in first: Import Your Existing Memory.
Managed relayer or self-hosted relayer?
The managed relayer covers WAL storage fees and SUI gas through its server wallets and configured gas sponsor. You pay no gas or WAL for relayed writes and run no infrastructure. In exchange, the shared service applies usage limits and carries no service level agreement (SLA).
A self-hosted relayer moves cost and operations to you:
- Funding: You fund the relayer deployment, including access to a wallet funded with WAL and SUI.
- Embeddings: You supply your own OpenAI-compatible API and credentials.
- Isolation: You can deploy your own contract with a separate package ID, Seal encryption keys, and hard data isolation.
- Availability: You control uptime instead of relying on a shared service without an SLA.
Start with the managed relayer if you want to integrate without running infrastructure. Self-host when you need your own instance, trust boundary, or availability guarantees.