R. S., Berlin writes: I've been renaming the built-in agents after my two cats (Fritz and Mirabel). My partner thinks this is stupid but I think it makes the app feel more mine. Who's right?
Your partner. Sorry.
We have strong opinions about this because we watched it happen to ourselves internally. For six weeks in the beta, everyone on the team named their agents after something personal — pets, ex-partners, childhood stuffed animals, one "Gandalf" that we kept pretending was ironic. It was great, briefly.
Then Mira on the engineering team made a routine called "ask Mittens to pull the staging logs" and someone else on the team forgot which cat was which agent and a deploy went sideways. Not a disaster, but the "wait which one is reading my Gmail?" problem is real. Agents have different tool scopes; if you've named them after animals you personally love, you will lose track, because your brain refuses to rank your pets by their permission levels.
Lime, Yuzu, Clem, and Cherry aren't the most charming names in the world. We picked them because:
- They're distinct enough to remember at a glance — four fruits, four colors, four roles.
- They describe the role without being bossy — Lime is curious, Yuzu is careful, Clem is eager, Cherry is patient.
- They're not tied to anyone's real life, so the app works the same whether you're the original owner of the vault or the new hire reading someone else's routines.
If you really can't help yourself, change the display name (Settings → Agents → Display) and leave the handle untouched. That way @clem still means Clem in every routine and memory file — but the chat header says "Mittens." Your brain wants the joke; your routines want the stable reference. Both can win.
But also — trust us — give it a month. It stops feeling impersonal faster than you think.
— The editors