Compaction | OpenAI API
Compaction Manage long-running conversations with server-side and standalone compaction. Overview To support long-running interactions, you can use compaction to reduce context size while preserving state needed for subsequent turns. Compaction helps you balance quality, cost, and latency as conversations grow. Server-side compaction You can enable server-side compaction in a Responses create requ
ゼãƒã‹ã‚‰å§‹ã‚ã‚‹Context Engineering
ã•ã‚‰ã«ã“ã‚ŒãŒ2ターン目ã€3ターン目ã¨ãƒ«ãƒ¼ãƒ—回数ãŒå¢—ãˆã‚‹ã”ã¨ã«90%ã«æ¼¸è¿‘ã—ã¾ã™ (ãªãŠã€Context Windowã®å®ŸåŠ¹åˆ¶é™ãŒã‚ã‚‹ãŸã‚ã€ã ã„ãŸã„85%安ã„ã¨ã„ã†ã‚¤ãƒ¡ãƒ¼ã‚¸ã‚’æŒã¤ã¨ã‚ˆã„ã§ã—ょã†)。ã‚ャッシュã¯ã‚¹ãƒ«ãƒ¼ãƒ—ットやレイテンシã®æ©æµã‚‚ã‚ã‚‹ãŸã‚ã€ç¾ä»£çš„ãªAI Agentã§ã¯å¿…é ˆã®ã‚‚ã®ã¨ãªã‚Šã¾ã™ã€‚ ã“ã‚Œã¯vLLMç‰ã§ã‚»ãƒ«ãƒ•ãƒ›ã‚¹ãƒ†ã‚£ãƒ³ã‚°ã—ãŸå ´åˆã‚‚åŒæ§˜ã§ã™ã€‚(ãªãŠã€ç´°ã‹ã„話ã§ã¯ã‚ã‚Šã¾ã™ãŒã€vLLMã§ã¯KVã‚ャッシュをオフãƒãƒ¼ãƒ‰ã—ãŸã„ã¨ãLMCacheを使ã†ã®ãŒãƒžã‚¹ãƒˆã§ã—ãŸãŒã€ã‚³ãƒã‚¯ã‚¿ãŒå°Žå…¥ã•ã‚ŒCPUオフãƒãƒ¼ãƒ‰ã«ã‚‚最近対応ã—ã¾ã—ãŸ) Context Engineeringã®ã¤ã‚‰ã„ã¨ã“ã‚ ã•ã¦ã€ã“ã®Cacheã®å®‰ã•ã‚’見るã“ã¨ã§Context Engineeringã§é¿ã‘ãŸã„ã“ã¨ãŒè¦‹ãˆã¦ãã¾ã™ã€‚ãã‚Œã¯ã‚³ãƒ³ãƒ†ã‚ストã®é€æ¬¡åœ§ç¸®ã§ã™ã€‚ãªãœã‹ã¨ã„ãˆã°ã€Context CachingãŒåŠ¹ã‹ãªããªã£ã¡ã‚ƒã†ã‹ã‚‰ã§
I Reverse Engineered ChatGPT's Memory System, and Here's What I Found! - Manthan
When I asked ChatGPT what it remembered about me, it listed 33 facts from my name and career goals to my current fitness routine. But how does it actually store and retrieve this information? And why does it feel so seamless? After extensive experimentation, I discovered that ChatGPT’s memory system is far simpler than I expected. No vector databases. No RAG over conversation history. Instead, it
Effective harnesses for long-running agents
Published Nov 26, 2025 Agents still face challenges working across many context windows. We looked to human engineers for inspiration in creating a more effective harness for long-running agents. As AI agents become more capable, developers are increasingly asking them to take on complex tasks requiring work that spans hours, or even days. However, getting agents to make consistent progress across
手触り感ã®ã‚ã‚‹Context Engineering - LayerX エンジニアブãƒã‚°
ã“ã¡ã‚‰ã¯LayerX AI エージェントブãƒã‚°ãƒªãƒ¬ãƒ¼2日目ã®è¨˜äº‹ã§ã™ï¼ˆ1日目ã®ponã•ã‚“ã®æ€’涛ã®TKG記事（not Tamago kake gohan）もãœã²ã”覧ãã ã•ã„）。 ã“ã‚“ã«ã¡ã¯ã€CEO室ã§AI Agent開発ã®PdMã‚’ã‚„ã£ã¦ã„ã‚‹Kenta Watanabeã§ã™ã€‚ è¿‘å¹´ã®LLM関連技術ã®æ€¥é€Ÿãªç™ºé”ã«ã‚ˆã‚Šã€è‡ªç¤¾ãƒ—ãƒãƒ€ã‚¯ãƒˆã®é–‹ç™ºã«LLMを活用ã™ã‚‹æ–¹ã‚‚増ãˆã¦ãã¦ã„ã‚‹ã®ã§ã¯ãªã„ã‹ã¨æ€ã„ã¾ã™ã€‚一方ã§ã€LLMã®ç¢ºçŽ‡çš„ãªæŒ¯ã‚‹èˆžã„ã‹ã‚‰ãƒ—ãƒãƒ€ã‚¯ã‚·ãƒ§ãƒ³ã§å®‰å®šç¨¼åƒã™ã‚‹æ©Ÿèƒ½ã‚„AI Agentã®é–‹ç™ºã«è‹¦æˆ¦ã—ã¦ã„る方もåŒæ™‚ã«å¤šã„ã®ã§ã¯ãªã„ã‹ã¨æ€ã„ã¾ã™ã€‚ ãã†ã„ã£ãŸä¸ã§ã€6æœˆé ƒã‹ã‚‰Context Engineeringã¨å‘¼ã°ã‚Œã‚‹LLMã‚’ã†ã¾ã稼åƒã•ã›ã‚‹ãŸã‚ã®æŠ€è¡“ãŒè©±é¡Œã«ãªã£ã¦ãã¾ã—ãŸã€‚Context Engineeringã¨ã„ã†ã‚ーワードãŒãƒã‚ºã‚Šå‡ºã—ãŸèµ·æºã‚„Context Engineering自体ã®è§£èª¬ã¯å„所
AIエージェントを支ãˆã‚‹æŠ€è¡“: コンテã‚ストエンジニアリングã®ç¾åœ¨åœ° - Algomatic Tech Blog
ã¯ã˜ã‚ã« ã“ã‚“ã«ã¡ã¯ã€‚Algomatic AI Transformation(AX) ã®sergicalsix（@sergicalsix）ã§ã™ã€‚ 本記事ã§ã¯å¤§è¦æ¨¡è¨€èªžãƒ¢ãƒ‡ãƒ«(LLM)を用ã„ãŸã‚¢ãƒ—リケーションãªã„ã—AIエージェントã®æ§‹ç¯‰ã«ãŠã„ã¦åˆ‡ã£ã¦ã‚‚切り離ã›ãªã„「コンテã‚ストエンジニアリングã€ã«ã¤ã„ã¦2025å¹´10月時点ã§ã®çŸ¥è¦‹ã‚’備忘録ã¨ã—ã¦ã¾ã¨ã‚ã¾ã™ã€‚ コンテã‚ストエンジニアリングã¨ã¯ コンテã‚ストエンジニアリングã¨ã¯ã€LLM ã«ä¸Žãˆã‚‹æƒ…å ±(コンテã‚スト)を制御ã™ã‚‹æŠ€è¡“ã§ã™ã€‚ コンテã‚ストエンジニアリングã¯ã‚ˆãプãƒãƒ³ãƒ—トエンジニアリングã¨å¯¾æ¯”ã•ã‚Œã¾ã™ã€‚ プãƒãƒ³ãƒ—トエンジニアリングã¨ã‚³ãƒ³ãƒ†ã‚ストエンジニアリング (Effective context engineering for AI agents) プãƒãƒ³ãƒ—トエンジニアリングã¯ã‚ãã¾ã§ç‰¹å®šã®ã‚¿ã‚¹ã‚¯ã«ç‰¹åŒ–ã—ãŸã‚¨ãƒ³ã‚¸ãƒ‹ã‚¢ãƒªãƒ³ã‚°æ‰‹æ³•ã§ã‚ã‚‹ã®ã«
Context Engineering for AI Agents: Lessons from Building Manus
Context Engineering for AI Agents: Lessons from Building Manus At the very beginning of the Manus project, my team and I faced a key decision: should we train an end-to-end agentic model using open-source foundations, or build an agent on top of the in-context learning abilities of frontier models? Back in my first decade in NLP, we didn't have the luxury of that choice. In the distant days of BER
ClaudeCodeã§Context Engineeringã«æ²¿ã£ãŸé–‹ç™ºãƒ•ãƒãƒ¼ã‚’探索ã™ã‚‹
ã¯ã˜ã‚㫠日々変化ã™ã‚‹AIã®ä¸–ç•Œã«ãŠã„ã¦ã€ã€ŒContext Engineeringã€ã¨ã„ã†è¨€è‘‰ãŒVibe Codingã«ç¶šãæ–°ãŸãªãƒˆãƒ¬ãƒ³ãƒ‰ã¨ãªã‚Šã¤ã¤ã‚ã‚Šã¾ã™ã€‚ トレンドã¨ãªã£ãŸãã£ã‹ã‘ã¯ã“ã®2人ã®ãƒ„イートã®ã‚ˆã†ã§ã™ã€‚ Vibe Codingã®æ™®åŠã«ã‚ˆã‚Šã€POC開発や趣味プãƒã‚¸ã‚§ã‚¯ãƒˆã«ãŠã„ã¦ã¯ã€äººé–“ãŒ1行もコードを書ã‹ãªãã¦ã‚‚å‹•ãã‚‚ã®ãŒä½œæˆã•ã‚Œã‚‹ã¨ã“ã‚ã¾ã§é€²åŒ–ã—ã¦ãã¦ã„ã¾ã™ã€‚åŠ ãˆã¦ã€ä¼æ¥ã«ãŠã‘るソフトウェア開発ã«ãŠã„ã¦ã‚‚ã€äººé–“ã®ä»‹å…¥ãªã—ã«AIãŒã‚³ãƒ¼ãƒ‰ã®ã»ã¨ã‚“ã©ã‚’開発ã§ãるケースも増ãˆã¦ãã¾ã—ãŸã€‚ ãŸã ã€å®Ÿéš›ã«AIãŒæ›¸ã„ãŸã‚³ãƒ¼ãƒ‰ã‚’人間ã®ãƒ¬ãƒ“ューãªã—ã«ãƒªãƒªãƒ¼ã‚¹ã§ãã‚‹ã‹ã¨ã„ã†ã¨ã€ãã“ã¾ã§ã«ã¯è‡³ã£ã¦ã„ãªã„ã®ãŒç¾çŠ¶ã§ã™ã€‚600人ã»ã©ã‚’対象ã«qodoãŒå®Ÿæ–½ã—ãŸãƒ¬ãƒãƒ¼ãƒˆã‚’確èªã—ãŸã¨ã“ã‚ã«ã‚ˆã‚‹ã¨ã€76%ã®ã‚¨ãƒ³ã‚¸ãƒ‹ã‚¢ãŒäººé–“ã®ãƒ¬ãƒ“ューãªã—ã«ãƒªãƒªãƒ¼ã‚¹ã™ã‚‹ã»ã©ã®ä¿¡é ¼æ„ŸãŒãªã„ã¨å›žç”ã—ã¦ã„るよã†ã§ã™ã€‚ 今後ã€ä¼æ¥ã«ãŠã‘
Context Engineering
TL;DRAgents need context to perform tasks. Context engineering is the art and science of filling the context window with just the right information at each step of an agent’s trajectory. In this post, we break down some common strategies — write, select, compress, and isolate — for context engineering by reviewing various popular agents and papers. We then explain how LangGraph is designed to supp
1
メンテナンス
ランã‚ング
ãŠçŸ¥ã‚‰ã›
リリースã€éšœå®³æƒ…å ±ãªã©ã®ã‚µãƒ¼ãƒ“スã®ãŠçŸ¥ã‚‰ã›
最新ã®äººæ°—エントリーã®é…ä¿¡
j次ã®ãƒ–ックマーク
kå‰ã®ãƒ–ックマーク
lã‚ã¨ã§èªã‚€
eコメント一覧を開ã
oページを開ã
https://x.com/i/status/2031798630177251514
https://x.com/i/status/2031416367610744960
New ETH Zurich Study Proves Your AI Coding Agents are Failing Because Your AGENTS.md Files are too Detailed
The Context Engineering Guide
Context Engineering in Manus
Architecting efficient context-aware multi-agent framework for production- Google Developers Blog
{{#tags}}- {{label}}
{{/tags}}