Context Engineering

Context Engineering is the discipline of deciding what an AI system sees at inference time — selecting, compressing, ordering, and refreshing the contents of the context window — so the model has exactly what it needs and little else.

It is broader than Prompt Engineering: it spans retrieval, Agent Memory, tool outputs, and history management, treating the window as a scarce budget to curate rather than fill. What you leave out matters as much as what you put in.

For builders it is the main lever on agent reliability and cost: irrelevant or contradictory context degrades reasoning and inflates token spend. The common failure is dumping everything available into the prompt, which buries the signal and triggers the lost-in-the-middle effect on Long Context inputs.

Topic: LLMs Related: Long Context Prompt Engineering Agent Memory RAG Agents

Recent Updates

• 2026-07-23: A production Android app built mostly by AI from detailed specs (Искусственный интеллект – AI, ANN и иные формы искусственного разума) · habr.com — Google · Figma · Microsoft
• 2026-07-23: A case study on using local LLMs for text-to-SQL in a WMS billing workflow (Искусственный интеллект – AI, ANN и иные формы искусственного разума) · habr.com — Аэросиб-С Ollama
• 2026-07-23: Whetuu: zero-config cross-shell prompt written in Zig (Hacker News) · yamafaktory.github.io
• 2026-07-23: ChatGPT Health can use connected medical records across conversations (OpenAI) · x.com — OpenAI · Apple Health
• 2026-07-23: Screenpipe adds local screen-and-audio capture as searchable memory for agents (Hacker News) · news.ycombinator.com — Hacker News Screenpipe · YC Tella · YouTube · Obsidian · OpenAI · Anthropic Louis Karpathy Garry gpt2 GPT3 · Parakeet · Whisper
• 2026-07-23: Claude-Thermos Keeps Claude Code Prompt Caches Warm (Hacker News) · github.com — Anthropic · GitHub
• 2026-07-24: Claude Cookbook roundup covers agent workflows, eval loops, memory, and managed agents (Hacker News) · platform.claude.com — Anthropic · OpenAI · Modal · Docker · Fable 5 · Opus 4.8
• 2026-07-24: Inference-time knowledge injection improves zero-shot delirium prediction in open-weight LLMs (cs.CL updates on arXiv.org) · arxiv.org — arXiv · Llama-3.1-8B · Llama-3.3-70b · GPT-5.2
• 2026-07-25: Vivix launches A1, a real-time interactive multimodal model with a unified streaming architecture (量子位) · qbitai.com — Vivix 量子位 · QbitAI henry A1 W1
• 2026-07-25: A debate on whether agents are already useful or still just getting started (量子位) · qbitai.com — QbitAI · Alibaba Kujing Technology Miaopai CCF · Tsinghua University Nanjing University Shanghai Jiao Tong University Tianjin University Shandong University Shanghai AI Laboratory Jin Lei Gao Yang Hao Jianyie Han Zhongyi Wen Ying Zhou Hao Zhang Hangfan Du Yanlong
• 2026-07-25: Anthropic’s Opus 5 is reported to match or beat Fable 5 on several benchmarks while Claude Code’s system prompt was cut by more than 80% (量子位) · qbitai.com — Anthropic · QbitAI · Cursor · OpenAI AlphaSchool 克雷西 am.will OmedTheVibeCoder Alex Ermolov Chetaslua Noema Matt Shumer Victor M Thariq · Opus 5 · Opus 4.8 · Fable 5 · GPT-5.6 Sol · GPT-5.6 · Kimi K3
• 2026-07-25: Claude Code trims system prompts as context engineering shifts (Hacker News) · claude.com — Claude Opus 5 · Claude Fable 5
• 2026-07-27: Wattage: A token-spend profiler and cost-regression gate for AI agents (Hacker News) · github.com — faizannraza
• 2026-07-27: Ant Group releases Ling-3.0-Flash, an agent-optimized hybrid inference model (量子位) · qbitai.com — Ant Group · OpenRouter Ling-3.0-Flash Ling-3.0
• 2026-07-27: Encoding Invisible Causation for Bridge Diagnostic Agents: Triple-Guided Retrieval-Augmented Fine-Tuning with QLoRA (cs.LG updates on arXiv.org) · arxiv.org
• 2026-07-27: Lost in Context: Addressing Context Anxiety in Large Language Models (cs.AI updates on arXiv.org) · arxiv.org
• 2026-07-27: Industrial Tokenization: Federated Architecture for LLM-Based Reasoning Over Heterogeneous Evidence (cs.AI updates on arXiv.org) · arxiv.org
• 2026-07-27: The Regression Tax: Decomposing Why Skills Help and Hurt LLM Agents (cs.AI updates on arXiv.org) · arxiv.org
• 2026-07-27: Learning What Matters: Supervising Sparse Attention Routing with Causal Evidence Sets (cs.CL updates on arXiv.org) · arxiv.org — Qwen2.5-3B · Gemma-2-9B
• 2026-07-27: Kimi-K3: Open-Source Frontier Model with Native Agentic and Coding Capabilities (Hacker News) · huggingface.co — Moonshot AI · Hugging Face · Kimi K3
• 2026-07-27: Kimi K3 Technical Report: 1M-Token Context with 2.5x Efficiency Over K2 (alphaXiv) · x.com — Kimi K3 · Kimi K2 · SigLIP
• 2026-07-28: How Context Attribution Handles What the Model Already Knows (cs.CL updates on arXiv.org) · arxiv.org
• 2026-07-28: Rebuilding legacy sites using AI without reading the code (Все статьи подряд / Искусственный интеллект / Хабр) · habr.com
• 2026-07-28: Small language models need fine-tuning to effectively use provided legal context (cs.CL updates on arXiv.org) · huggingface.co — Qwen3.5
• 2026-07-28: BHARATI: Morphology-Aware Tokenizers for Classical Indian Languages with Subword Fertility Analysis (cs.CL updates on arXiv.org) · arxiv.org — Santhosh Sivasubramani GPT-2 mBART-50

FAQ

What is Context Engineering?

Context Engineering is the discipline of selecting, compressing, ordering, and refreshing the information an AI system sees at inference time. GROUNDING tracks context windows, memory, tools, retrieval, and prompt architecture.

Which topic does Context Engineering belong to?

On the GROUNDING radar, Context Engineering is grouped under the LLMs topic.

Which concepts are related to Context Engineering?

Related concepts tracked by the radar include Long Context, Prompt Engineering, Agent Memory, RAG, Agents.