Story · arXiv
The Compaction Cliff in Long-Running AI Agent Memory (arXiv)
paper · Story page
A paper measures how many of an agent's safety rules survive repeated context compaction and proposes per-type retention policies.
In plain words
- Researchers found repeated shortening of an artificial intelligence assistant's working notes quickly erased exact safety rules.
- Ordinary activity records and safety instructions shared limited writing space, so both were shortened at the same rate.
- Claude Code preserved 53% of safety rules after one shortening and 10% after five.
- The proposed system labels each piece by purpose, preserves rules exactly, splits oversized topics, and retrieves stored information when needed.
- This could help developers keep long-running software assistants from forgetting safety limits while their working notes change.
Appeared in
- Compaction erases agent safety rules, and CLAUDE.md prose isn't a control
Aug 26, 2026 · lead story
Subscribe
Get the brief in your inbox
Pick daily, weekly, or both. Nothing is gated either way: every issue is on the site and in the feeds.
- Weekdays at 8:45am IST, one lead story and 6 to 9 items.
- Sundays, an argued synthesis rather than a recap.
- One click to leave, and quiet days say so in the subject line.