The Context Compaction Trap

Senior AI Engineer interview at Anthropic, and the interviewer asks:

“You swapped raw tool outputs for LLM-generated summaries in your coding agent. Prompt size dropped sharply. But cost per resolved issue didn’t move, and on some runs it went up. Where did the savings go, and how do you prove it from the traces?”

Don’t say: “The summarizer call itself costs tokens. Use a cheaper model for summarization.”