After consuming my whole usage window in a few minutes multiple times, I realised I needed a hook to do some checks for me and prompt me before submitting when cache had expired and content is too big, so that it can suggest me to "compact then send". It can also prevent sub-agents from spawning when above 80% usage window. Control is handed over to the user.<p>The first I built, based on hooks (run for 75 days), the biggest burst fell from 365 agent spawned to 27, and requests over 500k tokens of context from 14.5% to 0.6%. Lockouts still went up, because lots of parallel main sessions use up the window too. The strength is that it counts agents across every session on the machine.<p>I built <a href="https://github.com/rbartoli/agent-usage-guard" rel="nofollow">https://github.com/rbartoli/agent-usage-guard</a>, MIT-licensed, makes no network calls, needs Claude Code 2.1.287 or later.<p>I have also a post on this: <a href="https://rblab.com/what-stops-claude-code-burning-your-usage-window/" rel="nofollow">https://rblab.com/what-stops-claude-code-burning-your-usage-...</a>
interesting, I think that's a needed lever of control that's currently missing. Can you state against what plan you think the defaults make sense (pro / max 5x / max 20x)? A nice addition might be presets which are meaningful for each plan (i.e roughly scaled accordingly where it makes sense).