In the past I've use Termius (which is great), but I've never been thrilled about using it with coding agent CLI's. For typing in specific bash commands, it's fine, but agentic coding is much more verbose. Ideally we could use the phones native speech to text, but this doesn't work on a terminal UI - it needs a real, editable text field so it can insert and revise words as you speak, something a terminal UI can't provide.
'driftty' is a layer on top of 'ttyd' and gives you a real input box, letting you use speech to text, or quickly making edits to the text, then sending it to the agent when you're ready.
There's also a quick bar rather than a full keyboard, focused on commands and characters that are frequently used - '/' [tab] [shift tab] [enter] [arrow up] [arrow down] etc. - It's also got tmux specific functionality, letting you easily flip between buffers with one click.
This started as a quick side project and I've actually become surprised at how much I've used it, I've coded at least 90% of driftty on my phone (for better or for worse). It's still early, but I thought I'd throw it up there for HN to look at it. Note: the SSH gateway app is unauthed, the assumption is that you'll secure it (Cloudflare auth, caddy proxy, etc)