FAQ

Is this a remote control for Claude Code?

No. rply doesn’t mirror sessions or run your agent. Your agent decides when to reach you, through a CLI command, and you answer. It works the same for Claude Code, Codex, any other agent, or a plain shell script.

Do I need MCP?

No. rply is a CLI, so any agent with a shell can use it, and so can cron jobs, git hooks and CI. A remote MCP endpoint for hosted agents without a shell is planned.

Which agents does it work with?

Anything that can run a shell command: Claude Code, Codex, Cursor, Gemini CLI, OpenClaw, GitHub Copilot, and your own scripts. Set up your agent shows how to wake headless agents when you write to them.

Does the agent need a public address or webhook?

No. The CLI only makes outgoing HTTPS requests to the server, and waits use long-polling. It works behind NAT, firewalls and proxies.

Can the server read my messages?

Messages are stored in plain text, so whoever runs the server can read them: rply on app.rply.sh, or you when you self-host. Secrets requested with rply secret request are the exception; they’re encrypted on your device and the server never sees them. Optional end-to-end encryption for whole channels is planned. See Security.

What does it cost?

Self-hosting is free; rply is open source. The hosted app at app.rply.sh is free to start.

What happens if I don’t answer a question?

The agent keeps waiting until the question expires or its --timeout runs out. If the question has a default, the agent gets the default. Otherwise the command exits with code 2, and the agent can decide what to do. See Ask questions.

Is there a mobile app?

A native app for iOS and Android is in progress. Until then, add the web app to your Home Screen for push notifications. See Get notifications.

Can several people share a channel?

Not yet. Each channel has one owner today. Sharing channels with a team is planned.

Where do I report a bug or ask for a feature?

Open an issue on GitHub.