https://b.hatena.ne.jp/entry/s/arxiv.org/abs/2608.26263 Large Language Models (LLMs) increasingly act as autonomous agents executing complex, long-running procedural skills. Existing agent runtimes maintain execution by continually appending observation... https://b.hatena.ne.jp/yug1224/20260908#bookmark-4792417526658999202 2026-09-08T13:08:39Z https://b.hatena.ne.jp/ktykogm/20260908#bookmark-4792417526658999202 2026-09-08T11:21:56Z AI LLM Agent skills paper https://b.hatena.ne.jp/nguyen-oi/20260902#bookmark-4792417526658999202 履歴全追加じゃなくて状態のみ保持でトークン削減するのアプローチとして筋が良い 2026-09-02T02:44:31Z https://b.hatena.ne.jp/ZAORIKU/20260902#bookmark-4792417526658999202 2026-09-02T02:32:31Z https://b.hatena.ne.jp/gclover/20260901#bookmark-4792417526658999202 2026-09-01T14:06:39Z https://b.hatena.ne.jp/y_tag/20260901#bookmark-4792417526658999202 2026-09-01T13:42:34Z