Prompt lets Claude Code audit its own logs and find the 3 habits burning your usage limits
YashasGunderia · x · 2026-10-29
LimestoneHQ shared a prompt that makes Claude Code audit its own logs and surface the 3 habits burning your usage limit, at roughly 0.2% of a week's quota. The earlier version overcounted usage 2.4x and missed a third of it; this one fixes both, with a built-in check to run first.
Yashas Gunderia adds tuning tips: capable models need a clear objective, not more scaffolding. He recommends autocompact at 400k, turning off agent teams in favor of subagents, and asking Claude to audit and fix whatever is eating your usage.