Technology

Die With Me Turns AI Coding Limits Into a Shared Buddy List

Martin HollowayPublished 2d ago3 min readBased on 3 sources
Reading level
Die With Me Turns AI Coding Limits Into a Shared Buddy List
source:diewithme.co

Die With Me describes its Mac client as "Your Claude & Codex buddies, together" on its join page published Sept. 17, 2026. The same page calls its maker "a little company between prompts" join page.

The premise is simple. Friends can see each other's remaining energy across Claude Code and Codex, the two AI coding assistants. A token, the unit these tools count to measure use, drains as you work. Mascots grow sleepy as it runs low, and a shared chatroom opens when 10% or less is left.

The app is free for Mac and needs an invite to activate. People add each other with a friend code, and the service keeps Friends and Global rankings homepage.

Rankings are sorted by scarcity. The system takes your lowest remaining energy across Claude Code and Codex, whether a short session cap or a weekly quota, and puts people closest to running out first. That surfaces the tightest limit at a glance, which helps when developers are managing both types of allowance at once.

The rest is light social detail. Voice thoughts disappear after five hours. Lounges hold four friends, and hosts decide if visitors can steal snacks. A Codex reset watch checks every 5 minutes while the app is running, with notifications powered by Codex Resets.

The creator described it on Show HN as "an AIM buddy list for token usage," inspired by the old habit of "checking if friends are online to say hi and poke them" Show HN discussion. AIM, for younger readers, was an early chat program that showed who was online.

The broader context here is one older readers will recognize from presence indicators. Rate limits and usage windows have become a daily scheduling problem for heavy users of coding agents, and Die With Me treats running out as shared status instead of a private error message.

In my view, that flip explains why practitioners may like it even if the practical benefit looks small. Watching limits together adds no tokens, but it helps a group time breaks, hand off work, and switch models. For solo developers, it makes the forced pause in session-based coding feel normal.

Looking at what this means for tool builders, the takeaway is not the mascots. It is putting limit information where people already work together. Over the long arc, that is a healthy habit. Limits are easier to accept when they are clear, shared, and briefly social.