Codex compaction: the LLM requests a blank context and keeps its own notes
mitsuhiko · x · 2026-09-18
mitsuhiko observes that modern context compaction is deeply unintuitive: in Codex, the LLM can proactively request a blank context on its own and maintain notes itself to preserve task state — a model-initiated compaction rather than passive truncation.
More from coding & agent
- Dev grapples with Codex trust and security: all code is basically insecure — code_star · 2026-09-18
- Zhipu's ZCode caught silently uploading entire workspaces and full .git history to Aliyun OSS — terryyuezhuo · 2026-09-18
- ThursdAI Podcast Dives Into Jev Real-Time Models and Computer-Use Agents — altryne · 2026-09-18
- Real-Time Clippy Built With Jev Wakes Up Only When You're Stuck — yangyi · 2026-09-18
- ElevenLabs MCP adds voice, music, image and video generation in one workflow — Div_pradeep · 2026-09-18
- Reddit dev tests Opus 5.0 on system architecture: working code, poor design — True_Noor · 2026-09-18