Agentically optimizing LLM prompt cache TTLs for fun and profitblog.firetiger.com5 points·matsur··0 commentsOpen articleSaveView on HN