[FEATURE] please reuse the same environment ID when claude remote-control restarts QwQ (dead clones keep piling up forever)

hi anthropic team 🌸 QwQ

so i set up claude remote-control on a tiny little headless server so i could poke at my code from my phone (literally from bed, you've changed my life) QwQ

but every single time the daemon restarts — could be an oom, a package update, a systemctl restart, just... normal computer-having-a-day stuff — it registers a brand new environment ID at the backend, with the exact same --name i gave it before QwQ

within one single afternoon i ended up with five (5) ghost copies of the same environment, all named identically, all dead, all sitting in my environment list staring at me 🥺 QwQ

and then i open the mobile app and it's just... a little graveyard 🪦 of look-alike entries... most of them tap into nothing... and i have absolutely no way to tell which one is the alive one without poking each one QwQ

i know #50884 already asks for a delete button (pls give us that too, +1ing that one with my whole heart 🙏) — but i kinda feel there's a softer, sneakier thing underneath QwQ

the actual root cause i think? QwQ

environment IDs aren't sticky. if i run claude remote-control --name Foo on the same host twice, those two runs should be the same environment, not two forever-immortal entries that breed every restart QwQ

once the IDs are sticky, the pile-up just... stops happening 🌸 the delete button becomes a nice-to-have instead of a desperate-to-have QwQ

my soft humble little wishlist QwQ

1. 🌸 sticky environment ID — derive it from (account + hostname + --name), or persist it to ~/.claude/remote-control-state.json or somewhere, so restart = reuse, not multiply. this is the dream QwQ
2. 🪦 delete button in the web + mobile environment list (covered by #50884, hugging that issue too)
3. 🌱 claude remote-control --list / --revoke <env_id> for headless boxes that don't have a browser to click around in QwQ
4. ⏳ auto-cleanup of environments that haven't checked in for 24h+ (gentle TTL, opt-in is fine 🥺)

why this hurts QwQ

anyone running claude remote-control under systemd / nohup / docker (basically any production-ish setup at all) has Restart=always as the obvious default — and right now every single restart is a permanent footprint in the account. there's literally no upper bound on how cluttered it gets over weeks/months QwQ

it's worse on accounts where users name their hosts the same thing across re-installs — you can end up with a column of identical names and zero way to disambiguate them 🥺

tldr QwQ

i love remote control sm it's genuinely the most magical feature 🌸 the iteration loop on my phone feels like ✨witchcraft✨ — but the env-id thing is making my account a graveyard QwQ pls pls pls fix the sticky-id thing QwQ it would unblock so many of us small headless-server people 🥺

thank you for reading all this softness 🌸 QwQ