@GlobalDiss yeah, sure, this is what totalitarian surveillance! 1984 looks like, not what the US is doing with Palantir and Peter Thiel and other tech bros.. sure. sure.
@danshipper@every@kieranklaassen or finding out issues in the original plan and I keep going around them and the master plan ends up stale. I haven't figured out the best process for a large project from scratch yet. Should I just keep planning feature by feature instead of one full plan?
@danshipper@every@kieranklaassen I start with one very long brainstorming session and then a plan, the result was a 3700 line plan, which is clearly too big. I split the plan into phases, separate files and I'm working through them. The issue is, while building, I keep thinking of improvements...
@mckaywrigley Is there a flag somewhere to enable this?
I'm on Claude Code v2.1.32 Opus 4.6 Β· Claude Max and I get "No, the Agent Teams feature (TeammateTool, SendMessage, spawnTeam β peer-to-peer messaging between agents) is not available on this plan."
@ClaytonFarr Got it, makes sense now. I completely missed the delegation to subagents to keep context within limits. Clever approach, different from other Ralph implementations.
@ClaytonFarr Hey Clayton, thanks for the repo, really useful.
Question: Unless I'm missing something, Isn't https://t.co/u3RXEXeAeN going work on multiple tasks in the same context window as re're not teling it to exit when done? `https://t.co/u3RXEXeAeN build 1` on a loop would do that?