LLM Context Engineering
Parent: AI Agent Ecosystems · researched 2026-06-01T02:26:46.432Z· 27 sources · 8 concepts · skill ai-agent-engineering
AI & agent-engineering family ROUTER. Split into: ai-agents-orchestration (agent frameworks, multi-agent, memory, planning, guardrails, coding/GUI agents, autonomous loops, eval); ai-rag-retrieval (RA
ai-agent-engineering
- AI & agent-engineering family ROUTER. Split into: ai-agents-orchestration (agent frameworks, multi-agent, memory, planning, guardrails, coding/GUI agents, autonomous loops, eval); ai-rag-retrieval (RAG, iterative retrieval, vector/graph datastores); ai-llm-model-layer (training, fine-tuning, alignment/RLHF, compression, inference serving, transformer/multimodal architecture, model selection, observability); ai-mcp-sdk-prompting (MCP servers/builder, Anthropic SDK, prompt engineering, context engineering, LLM frameworks, tool-search, prompt lookup). Route to the matching sub-hub. [source]
- This hub routes to on-demand reference files under references/. See each spoke for depth. [source]
Children
- Context-as-managed-resource / RAM framing (Karpathy LLM-as-OS) (frontier)
- Context rot / lost-in-the-middle / effective-vs-advertised length (NoLiMa, RULER) (frontier)
- Compaction & context editing (auto-compact ~95%, Claude memory tool) (frontier)
- Just-in-time / agentic retrieval vs pre-loading (frontier)
- External memory & filesystem-as-memory (recoverable compression) (frontier)
- Multi-agent context (shared traces vs isolation: Cognition vs Anthropic) (frontier)
- KV-cache-aware layout (stable prefix, append-only, mask tools) (frontier)
- Failure modes (poisoning/distraction/confusion/clash - Breunig) (frontier)
Frontier under this node: Compaction & context editing (auto-compact ~95%, Claude memory tool), Context rot / lost-in-the-middle / effective-vs-advertised length (NoLiMa, RULER), Context-as-managed-resource / RAM framing (Karpathy LLM-as-OS), External memory & filesystem-as-memory (recoverable compression), Failure modes (poisoning/distraction/confusion/clash - Breunig), Just-in-time / agentic retrieval vs pre-loading, KV-cache-aware layout (stable prefix, append-only, mask tools), Multi-agent context (shared traces vs isolation: Cognition vs Anthropic)