--- title: "Context Layer Glossary: Enterprise AI Terms" url: "https://atlan.com/context-layer/glossary/" description: "Canonical definitions for every term in the context layer for AI - from context engineering and RAG to agent memory, MCP, and knowledge graphs." keywords: "context layer glossary, context engineering, context window, RAG, GraphRAG, knowledge graph, MCP, agent memory, context rot, context poisoning, semantic layer" --- > Atlan is hosting Context Conference, bringing together the leaders and builders at the frontier of giving AI the context it needs to understand their business. It runs online on October 28, 2026, from 11:00 AM to 2:00 PM ET. Atlan co-founder Prukalpa Sankar opens and closes the day. Leaders from AstraZeneca, BNY and Verizon share why they invest in context and what they get from it. Registrants get early access to The AI Context Gap, a new study from MIT Technology Review Insights. Register: https://atlan.com/context-conference/ Atlan's context layer glossary: canonical definitions for every term in Atlan's context layer architecture. 26 terms, from context engineering and RAG to agent memory, MCP, and knowledge graphs, each with a short definition, related terms, and a link to a longer Atlan guide or blog post. Last updated June 5, 2026 (as shown on the page). Hub: [The context layer](https://atlan.com/context-layer/) ## A **AI Debt** - Accumulated cost from shortcuts in semantic infrastructure - grows like financial debt. See also: Context Layer, Decision Intelligence. Blog: [Closing the context gap (the cost of unmanaged AI)](https://atlan.com/know/closing-the-context-gap/) **Agent Memory (Episodic)** - Memory of specific past events and interactions. See also: Agent Memory (Semantic), Agent Memory (Procedural), Context Management. Guide: [Episodic memory for AI agents, explained](https://atlan.com/know/episodic-memory-ai-agents/) **Agent Memory (Procedural)** - Memory of how to perform specific tasks and workflows. See also: Agent Memory (Episodic), Agent Memory (Semantic). Guide: [Semantic vs. procedural memory for AI agents](https://atlan.com/know/semantic-memory-vs-procedural-memory-ai-agents/) **Agent Memory (Semantic)** - Memory of general facts, definitions, and concepts. See also: Agent Memory (Episodic), Agent Memory (Procedural), Knowledge Graph. Guide: [Semantic vs. procedural memory for AI agents](https://atlan.com/know/semantic-memory-vs-procedural-memory-ai-agents/) ## C **Context Assembly** - The process of building the context window on the fly for each specific request. See also: Context Engineering, Context Window, Context Packs. Guide: [Dynamic context vs. static context](https://atlan.com/know/dynamic-context-vs-static-context/) **Context Clash** - Conflicting information from multiple sources present simultaneously in the context window. See also: Context Confusion, Context Poisoning. Blog: [RAG accuracy problems and how to fix them](https://atlan.com/know/rag-accuracy-problems/) **Context Confusion** - Performance degradation caused by too many tools or instructions competing for the model's attention. See also: Context Clash, Context Management. Blog: [Agent harness failures and anti-patterns](https://atlan.com/know/agent-harness-failures-anti-patterns/) **Context Engineering** - Designing the entire information environment in which an AI model operates. See also: Context Assembly, Context Layer, Prompt Engineering. Guide: [What is context engineering? The complete guide](https://atlan.com/know/what-is-context-engineering/) **Context Layer** - The architectural system that curates and delivers context to AI agents - composed of Semantics, Operational State, and Provenance. See also: Semantic Layer, Knowledge Graph, Context Engineering. Blog: [What is the context layer for AI?](https://atlan.com/know/what-is-context-layer/) **Context Management** - Keeping the context window current and relevant through summarisation, trimming, offloading, and ordering. See also: Context Rot, Context Window. Blog: [Context management at enterprise scale](https://atlan.com/know/context-management-cdo-enterprise-scale/) **Context Packs** - Pre-assembled, reusable bundles of context built for specific workflows. See also: Context Assembly, Context Engineering. Guide: [How to structure context for LLM applications](https://atlan.com/know/how-to-structure-context-for-llm-applications/) **Context Poisoning** - Wrong or manipulated information entering the context and being treated as fact. See also: Context Clash, Context Rot, Guardian Agents. Blog: [Context poisoning in AI systems, explained](https://atlan.com/know/context-poisoning/) **Context Rot** - Gradual degradation as the context window fills with low-signal content over a long session. See also: Context Management, Context Poisoning. Blog: [Context drift in AI agents](https://atlan.com/know/context-drift-ai-agents/) **Context Window** - The AI model's working memory - everything it can see right now. See also: Context Management, Context Engineering, Context Assembly. Guide: [Memory layer vs. context window](https://atlan.com/know/memory-layer-vs-context-window/) ## D **Decision Intelligence** - Embedding decision intent, constraints, and success metrics into agentic systems as explicit context. See also: Guardian Agents, Context Layer. Guide: [What are decision traces for AI agents?](https://atlan.com/know/what-are-decision-traces-for-ai-agents/) ## G **GraphRAG** - RAG enhanced with knowledge graph traversal - retrieves relationships between facts, not just similar text. See also: RAG, Knowledge Graph. Guide: [What is GraphRAG?](https://atlan.com/know/what-is-graphrag/) **Guardian Agents** - Independent AI agents that monitor and can block other agents' behaviour at machine speed. See also: Human-on-the-Loop, Decision Intelligence. Guide: [AI agent risks and guardrails](https://atlan.com/know/ai-agent-risks-guardrails/) ## H **Human-in-the-Loop** - Human approval required before each agent action. See also: Human-on-the-Loop, Guardian Agents. Guide: [AI agent risks and guardrails](https://atlan.com/know/ai-agent-risks-guardrails/) **Human-on-the-Loop** - Human sets intent upfront; agent acts autonomously; human monitors outcomes. See also: Human-in-the-Loop, Guardian Agents. Guide: [Autonomous agents vs. copilots](https://atlan.com/know/autonomous-agents-vs-copilots/) ## K **Knowledge Graph** - A database of entities and their relationships - enabling agents to reason across connected facts. See also: GraphRAG, Context Layer, Agent Memory (Semantic). Blog: [What is a knowledge graph?](https://atlan.com/know/what-is-a-knowledge-graph/) ## M **MCP** - Model Context Protocol - a universal standard (USB-C for AI) for connecting models to external tools and data sources. See also: Context Layer, Context Assembly. Guide: [What is the Model Context Protocol (MCP)?](https://atlan.com/know/what-is-model-context-protocol/) ## O **Operational State** - Everything that is currently true - world state, execution state, environment state. See also: Context Layer, Provenance. Guide: [What is agentic AI?](https://atlan.com/know/what-is-agentic-ai/) ## P **Prompt Engineering** - Crafting better instructions and questions to improve model outputs. See also: Context Engineering. Blog: [What is prompt engineering?](https://atlan.com/know/what-is-prompt-engineering/) **Provenance** - The complete recorded history of where data came from, what decisions were made, and what actions were taken. See also: Context Layer, Operational State. Blog: [Data lineage tracking, explained](https://atlan.com/know/data-lineage-tracking/) ## R **RAG** - Retrieve-then-generate - look up relevant documents before answering, rather than relying on training data alone. See also: GraphRAG, Knowledge Graph. Guide: [What is RAG (retrieval-augmented generation)?](https://atlan.com/know/what-is-rag/) ## S **Semantic Layer** - See Context Layer. Blog: [Context layer vs. semantic layer](https://atlan.com/know/context-layer-vs-semantic-layer/)