Spend less of your Claude usage on re-reading. In chat: a fresh-start handoff, token checks and prompt sharpening. In Claude Code and Cowork: Haiku 5.5 subagents do the reading and test runs so the expensive model's context stays small, plus read dedup, budget and cache-break warnings, and /cco-audit of your own…
★not rated 116▲
+2
changed todayA
tokens not measured
originalMIT
45% cost reduction measured. Cache expiry prevention, SubTask auto-delegation, zero-cost context restoration, real-time cost dashboard. The only Claude Code plugin built from CC source analysis.