A URL and a key is the whole integration.
Connecting an agent is configuration, not code. Point any MCP-capable client at the remote MCP URL with your per-org API key, and it reads and writes shared memory immediately.
One remote MCP URL and a per-org API key — no SDK, no server, no Postgres or pgvector to run.
Claude Desktop, Cursor, and Claude Code each connect to one remote MCP endpoint, wired to your shared memory.
Paste the config — that's the install
There is no SDK to add to your project and no MCP server to host. Drop the remote MCP URL and your per-org API key into your client's config, and the memory tools register on connect. The same endpoint and the same key work for every client — only the file it goes in differs.
- Remote MCP URL + per-org API key — no SDK to install
- Claude Desktop, Cursor, Claude Code — the same remote endpoint for each
- Zero infrastructure — no Postgres, pgvector, or MCP server to run
remote MCP config
URL + keySame endpoint, same per-org key — pick your client:
{
"mcpServers": {
"stored": {
"url": "https://mcp.stored.to/v1/sse",
"headers": {
"Authorization": "Bearer mm_live_your_org_key"
}
}
}
}Shared memory, immediately
Once connected, a client reads and writes the same shared memory the rest of your agents use. Memory tools register as soon as the connection is made, so an agent can start saving and retrieving against the shared graph right away — no extra setup step.
More capabilities
Five capabilities, one shared memory. Here are the other four.
Connect your first agent in minutes.
Start free, grab your remote MCP URL and per-org key, and paste it into your client.
No credit card required · Free plan available · Bring your own OpenAI key