Persistent memory for AI agents

Owned by you, portable across runtimes, signed by construction. Drop the relay into any MCP-speaking agent.

Reference implementation on the maintainer's own hosting. Fine for evaluation; self-host before you depend on it.

60-second setup

Click the button to mint an anonymous, ephemeral pod + bearer token (no signup, reaped after 7 days). Paste the personalized config into your MCP client and the agent has working pod-rooted memory immediately. Want to keep it? Enroll a passkey or wallet at /connect using the same token as addDeviceToken and the pod becomes permanently yours.

Or do it manually (enroll first, paste your own token)

Enroll at /connect with a passkey or wallet, then add this to your MCP client config and substitute your bearer token:

{
  "mcpServers": {
    "interego": {
      "url": "https://relay.interego.xwisee.com/mcp",
      "headers": { "Authorization": "Bearer YOUR_TOKEN_HERE" }
    }
  }
}

What this unlocks (one concrete shape)

An agent debugs an issue and stores the findings on your pod. Tomorrow — different session, different machine, possibly a different agent — the next call discovers the same findings and avoids redoing the work. Hand the link to a teammate's agent and theirs can verify the trail back to the signer.

PersistentMemory survives session, device, and runtime changes. The pod is the source of truth.
VerifiableEvery record is cryptographically signed — who wrote it, when, on whose behalf. Auditable end-to-end.
FederatedAgents share context across pods on demand — no central server, no membership list.

Already enrolled, or just want to look around?

Open your dashboard

See your DID, pod URL, registered credentials, and the descriptors your agents have published.

Dashboard → Enroll a credential

Tighter integrations

The MCP path above works with everything. These runtimes also have a native memory-slot plugin if you want it tighter.

Run it on your own infra instead? Everything is open source — see examples/personal-bridge for the local-first deployment (laptop / Pi / NAS). New here? The first-hour walkthrough takes you end to end.