Set up a Matrix-side assistant for email and reminders

A developer linked a self-hosted Matrix bot to read-only email, memory files, and reminders.

Done with OpenClaw· Advanced· Oct 2026

What they did
They had Kimi create a local Matrix server and assemble a small agent loop inspired by Claw. The system ran inside bubblewrap, used separate identity and memory files, and could inspect email without ever seeing login secrets. Incoming Matrix posts were used to trigger new turns, while a heartbeat process handled task progress, memory consolidation, and reminders.
How it went
The setup ran using Kimi together with GLM5 and was reachable from a phone, a web app, and through Claude, Codex, or Kimi CLI with a small MCP wrapper. No statistics were shared, and the author said the code may be released later.
Tips
Sandboxing with bubblewrap is not a complete defense against hostile input, and the agent still had web access. The author planned to move the system into a VM, so test carefully before connecting it to anything important.

Try it yourself with OpenClaw

Help me build a personal assistant that I can talk to through [Matrix or another messaging app]. Let it check my email at [email provider] in read-only mode, keep memory files for my preferences and open tasks, and set reminders. Before it sends anything, deletes anything, submits anything, or makes a payment, stop and ask me to approve the action. Start by sending a test message that summarizes my unread email from today.

Fill in the [brackets]. Agents can make mistakes, so review anything before it is sent, paid or deleted.

Read the original on news.ycombinator.com ↗

Get 3 agent uses like this every morning

With the day's top AI news. Free, and one click to unsubscribe.