Hi HN, I'm Joshua. rply is an open-source inbox between you and your AI agents.
Rply is not meant to be remote control for sessions. Instead you get a control center to communicate with your remote agents, have them send you notifications, ask questions, share progress and expose full UIs for you to monitor progress.
I built it for an SEO agent I run unattended on a VM. I initially glued it together with ntfy for push notifications, an exposed site for seeing status and a cron monitored inbox for waking the agent. This worked but I kept setting it up for new projects and it became unwieldy.
Rply now is much simpler. The agent gets a cli, you get a simple container to host on the same VM or just connect to the hosted version for free.
Your agent sends cli commands and sets them up to wake it when you send messages. It can send questions, files to show you it's work and ask for secrets and receive them without ever seeing them.
What I also love is that it can set up apps for a channel which get their data from your VM. You can view your stats, see recent work, monitor the agents, etc.
ntfy is an amazing project and inspired this heavily. But what was missing is the back and forth communication with the agent, formatted questions, artifacts and bringing all that in one place.
Rply strives to be as simple as ntfy while being built for agents.
It is still an early project but it's working well for a lot of projects already. I would love to hear your feedback, open to any critique or improvements you have!
Quickest way to get started is to send your agent the prompt on https://rply.sh
joshuaknauber•50m ago
Rply is not meant to be remote control for sessions. Instead you get a control center to communicate with your remote agents, have them send you notifications, ask questions, share progress and expose full UIs for you to monitor progress.
I built it for an SEO agent I run unattended on a VM. I initially glued it together with ntfy for push notifications, an exposed site for seeing status and a cron monitored inbox for waking the agent. This worked but I kept setting it up for new projects and it became unwieldy.
Rply now is much simpler. The agent gets a cli, you get a simple container to host on the same VM or just connect to the hosted version for free. Your agent sends cli commands and sets them up to wake it when you send messages. It can send questions, files to show you it's work and ask for secrets and receive them without ever seeing them. What I also love is that it can set up apps for a channel which get their data from your VM. You can view your stats, see recent work, monitor the agents, etc.
ntfy is an amazing project and inspired this heavily. But what was missing is the back and forth communication with the agent, formatted questions, artifacts and bringing all that in one place. Rply strives to be as simple as ntfy while being built for agents.
It is still an early project but it's working well for a lot of projects already. I would love to hear your feedback, open to any critique or improvements you have!
Quickest way to get started is to send your agent the prompt on https://rply.sh