KAIST's WikiTrace: evidence-linked memory fixes agent "memory-induced degradation"
dair_ai · x · 2026-11-11
KAIST researchers show an agent's memory can make it fail tasks it would otherwise solve — they call it memory-induced degradation. The cause: lessons are inferred from runs labeled only success/failure and stored without the run itself, so wrong lessons can't be verified. They introduce WikiTrace, which links each lesson to a summary of why the run succeeded/failed and the raw trajectory; when an insight is vague, conflicting, or from a failed run, the agent follows the links with terminal tools before acting. Across ALFWorld, ScienceWorld, PlanCraft, AppWorld, and tau3-bench retail, it scores 10.7 points above the agent without memory. Code: github.com/yukyeonglee/wikitrace.
Related event: KAIST Finds Agent Memory Can Hurt Performance, Proposes WikiTrace Fix(2 posts)→
More from coding & agent
- Using a third-party inference provider with your coding harness hands it shell access to your machine — TheZachMueller · 2026-11-11
- Splash One: flight combat sandbox built with Opus 5.5 in PlayCanvas — willeastcott · 2026-11-11
- Next.js 16.5 delivers 3.3x faster builds for a leading AI lab — cramforce · 2026-11-11
- Plurai launches free agent simulation testing inside Claude Code and Cursor — PrajwalTomar_ · 2026-11-11
- PlayCanvas engine repo hits 10,000th issue/PR — and it's a Claude-authored merge — willeastcott · 2026-11-11
- AI-built game nerfed 40% after Claude playtest bot beat it in 6 minutes — builderjaydub · 2026-11-11