# Agent-Controlled Forgetting Cuts a Debugging Session's Prompt Tokens From 912,000 to 232,000 Without Retraining

A separate paper finds that a widely cited gain from encoder-only healing in SVD KV-cache compression disappears once each method gets its own tuned learning rate.

- Published: 2026-10-09T06:20:56.057Z
- Canonical: https://polylog.news/ai/2026-10-09/agent-controlled-forgetting-cuts-a-debugging-session-s-promp
- Publisher: Polylog (AI desk)
- Section: tech
- Sources: [arXiv cs.AI](https://arxiv.org/abs/2610.10590), [arXiv cs.LG](https://arxiv.org/abs/2610.10552), [Polylog editors](https://polylog.news), [OpenAI News](https://openai.com/index/legalon-halves-codex-costs)

Tool-using agents keep large tool outputs in context long after they have used them. Jan-Peter Franke's agent-controlled forgetting lets the acting model choose earlier tool results and replace each one with a short note at its original pos…

This story is for subscribers. Read it in full at https://polylog.news/ai/2026-10-09/agent-controlled-forgetting-cuts-a-debugging-session-s-promp (subscription information: https://polylog.news/pricing).