# Glossary Terms

* [1M Token Context Window](/okf/en/glossary/1m-token-context-window.md) - A 1 Million Token Context Window is a large language models ability to process and retain information from a document or conversation of up to 1 million tokens.
* [Agent Economics](/okf/en/glossary/agent-economics.md) - Agent Economics refers to the cost structure, efficiency logic, and economic trade-offs involved in operating AI agents in production systems. Unlike traditiona
* [Agent Governance](/okf/en/glossary/agent-governance.md) - Frameworks and tools for monitoring, controlling, and ensuring compliance of autonomous AI agents in enterprise deployment.
* [Agent Handoff](/okf/en/glossary/agent-handoff.md) - Agent handoff is the structured transfer of an active task, along with its full context and intermediate state, from one AI agent to another within a multi-agen
* [Agent HQ](/okf/en/glossary/github-agent-hq.md) - GitHub's multi-agent orchestration platform allowing developers to switch between Claude, Codex, and Copilot agents in a unified interface. Transforms GitHub fr
* [Agent Observability](/okf/en/glossary/agent-observability.md) - Agent Observability refers to the capability to monitor, measure, and understand the behavior, state, and decision-making processes of AI agents in real time. U
* [Agent Permission Profiles](/okf/en/glossary/agent-permission-profiles.md) - Agent Permission Profiles are reusable permission bundles that define what an AI agent is allowed to do inside its runtime environment. Instead of giving every 
* [Agent Pull Request](/okf/en/glossary/agent-pull-request.md) - An Agent Pull Request (Agent PR) describes the end-to-end process in which an AI coding agent — such as Claude Code, OpenAI Codex, or similar systems — autonomo
* [Agent Runtime](/okf/en/glossary/agent-runtime.md) - An agent runtime is the execution environment where AI agents plan work, call tools, read data, store intermediate state, and interact with external systems. It
* [Agent Runtime Architecture](/okf/en/glossary/agent-runtime-architecture.md) - Agent runtime architecture refers to the technical execution environment in which AI agents process tasks, invoke tools, and manage state. It is the layer betwe
* [Agent Teams](/okf/en/glossary/agent-teams.md) - Agent Teams is a feature that enables the parallel coordination of multiple AI agents working together on complex tasks. Instead of a single agent working seque
* [Agent Tool Surface](/okf/en/glossary/agent-tool-surface.md) - An agent tool surface is the complete set of tools, functions, and interfaces an AI agent is able to call at runtime. It describes not how any single tool is wi
* [Agent Trust Boundary](/okf/en/glossary/agent-trust-boundary.md) - An agent trust boundary is the explicit security line that defines which information, files, tools and outputs an AI agent is allowed to trust. In traditional s
* [Agent-Accessible APIs](/okf/en/glossary/agent-accessible-apis.md) - Agent-Accessible APIs are interfaces intentionally designed for autonomous AI agents, not just human developers. The foundation is machine readability: explicit
* [Agent-to-Agent Protocol (A2A)](/okf/en/glossary/agent-to-agent-protocol.md) - Agent-to-Agent Protocol (A2A) is a agentic AI concept in modern AI systems that enables autonomous agent capabilities. It plays a key role in enterprise AI depl
* [Agentic AI](/okf/en/glossary/agentic-ai.md) - AI systems that autonomously reason, plan, and execute multi-step tasks to achieve specific goals, rather than just generating text.
* [Agentic AI Security](/okf/en/glossary/agentic-ai-security.md) - Agentic AI Security refers to the use of AI agents to proactively identify, assess, and mitigate security risks within AI systems and broader IT environments. I
* [Agentic Commerce Protocol](/okf/en/glossary/agentic-commerce-protocol.md) - An open standard released by OpenAI and Stripe in October 2025 that enables AI agents to execute secure purchase transactions on behalf of users.
* [Agentic Compute](/okf/en/glossary/agentic-compute.md) - Agentic Compute describes the full execution load created when AI agents do more than generate a single answer and instead carry out multi-step work on their ow
* [Agentic Engineering](/okf/en/glossary/agentic-engineering.md) - Agentic Engineering is a structured software development approach where AI agents are integrated into the delivery process as controlled contributors, not treat
* [Agentic IDE](/okf/en/glossary/agentic-ide.md) - An agentic IDE is a development environment with an autonomous AI agent built into its core, able to carry out multi-step programming tasks on its own — writing
* [Agentic Payments](/okf/en/glossary/agentic-payments.md) - Agentic payments are the capability of an autonomous AI agent to initiate, authorize, and complete a payment on a user's behalf. Unlike conventional online chec
* [Agentic Product Feed](/okf/en/glossary/agentic-product-feed.md) - An agentic product feed is a structured stream of product data engineered specifically so autonomous AI agents — such as shopping assistants inside ChatGPT or o
* [Agentic Reasoning](/okf/en/glossary/agentic-reasoning.md) - Extended thinking for autonomous agents – the ability to explicitly "think" before taking actions. Claude Opus 4.5, GPT-5.2, and Gemini 3 Pro use additional com
* [Agentic Systems](/okf/en/glossary/agentic-systems.md) - Agentic Systems are AI systems that independently make decisions and take actions to solve complex tasks. Unlike traditional AI systems that only react to input
* [Agentic Workflow](/okf/en/glossary/agentic-workflow.md) - An iterative process where AI agents break down tasks, execute them, and self-correct through loops to ensure high-quality outcomes.
* [AGENTS.md](/okf/en/glossary/agents-md-file.md) - A markdown file placed in a code repository that provides structured instructions, conventions, and context to AI coding agents, improving their task completion
* [AI Agent Capacity Planning](/okf/en/glossary/ai-agent-capacity-planning.md) - AI agent capacity planning is the structured planning of compute, API quotas, concurrency, queues, budgets and fallbacks for production AI agents. Unlike classi
* [AI Agent Control Plane](/okf/en/glossary/ai-agent-control-plane.md) - An AI agent control plane is the operating and governance layer that plans, authorizes, monitors, and constrains AI agents. While the model proposes the next ac
* [AI Agent Framework](/okf/en/glossary/ai-agent-framework.md) - An AI agent framework is the software foundation developers use to build, run, and operate autonomous AI agents. It packages the recurring building blocks of an
* [AI Agent Governance](/okf/en/glossary/ai-agent-governance.md) - AI agent governance is the set of rules, controls, and responsibilities that lets organizations run AI agents safely, transparently, and in line with business g
* [AI Agent Identity](/okf/en/glossary/ai-agent-identity.md) - AI agent identity is the unique, verifiable identity an autonomous AI agent uses to authenticate itself to systems, APIs, and other agents. Unlike a human user 
* [AI Agent Infrastructure](/okf/en/glossary/ai-agent-infrastructure.md) - AI agent infrastructure is the technical layer that lets AI agents move from chat-style assistance to controlled execution. It includes model access, tool and A
* [AI Agent Operations](/okf/en/glossary/ai-agent-operations.md) - AI Agent Operations is the operating discipline for running AI agents reliably, safely, and economically after the prototype stage. It covers session and task m
* [AI Agent Optimization](/okf/en/glossary/ai-agent-optimization.md) - The practice of systematically improving AI agent performance through structured instructions, context management, and feedback loops.
* [AI Agent Permissions](/okf/en/glossary/ai-agent-permissions.md) - AI Agent Permissions are the explicit rights an AI agent receives across software systems, data sources, tools, and business workflows. A normal chatbot mainly 
* [AI Agent Security](/okf/en/glossary/ai-agent-security.md) - AI Agent Security is the security architecture for AI agents that do more than generate text. These systems can call tools, change files, run code, use APIs, in
* [AI Agent Workflow](/okf/en/glossary/ai-agent-workflow.md) - A structured sequence of tasks performed by an AI agent to achieve a business outcome, involving reasoning, decision-making, and adaptive responses.
* [AI Agents in Production](/okf/en/glossary/ai-agents-in-production.md) - AI Agents in Production refers to the deployment and operationalization of AI agents in real-world environments to automate tasks, assist users, and interact wi
* [AI Autonomy in Enterprises](/okf/en/glossary/ai-autonomy-in-enterprises.md) - AI Autonomy in Enterprises describes the extent to which AI systems can independently perform tasks, make decisions, and manage processes within an organization
* [AI Bill of Materials (AIBOM)](/okf/en/glossary/ai-bill-of-materials.md) - An AI Bill of Materials (AIBOM) is a machine-readable inventory of every component that makes up an AI system: the models and their weights, the training and fi
* [AI Code Review Gate](/okf/en/glossary/ai-code-review-gate.md) - An AI code review gate is an automated quality control checkpoint embedded in a CI/CD pipeline that uses an independent AI model to evaluate code changes before
* [AI Code Security Review](/okf/en/glossary/ai-code-security-review.md) - AI code security review is the structured security assessment of code produced with AI coding tools, autonomous agents, or automated development workflows. It c
* [AI Coding Agent](/okf/en/glossary/ai-coding-agent-definition.md) - An AI system designed to automate software development tasks including code generation, bug fixing, feature implementation, and code migrations. Unlike simple c
* [AI Coding Agent Guardrails](/okf/en/glossary/ai-coding-agent-guardrails.md) - AI coding agent guardrails are the technical and organizational controls that define what an AI coding agent may do inside a software development environment, w
* [AI Coding Agents](/okf/en/glossary/ai-coding-agents.md) - AI Coding Agents are autonomous or semi-autonomous AI systems that perform software development tasks independently or in collaboration with human developers. U
* [AI Coding Assistants](/okf/en/glossary/ai-coding-assistants.md) - AI Coding Assistants are tools that use large language models to help developers write, debug, and understand code. They provide code completion, generation, ex
* [AI Control Risks](/okf/en/glossary/ai-control-risks.md) - Challenges of maintaining human oversight over increasingly capable AI systems. Major theme at WEF 2026 governance discussions.
* [AI Cost Optimization](/okf/en/glossary/ai-cost-optimization.md) - AI Cost Optimization encompasses strategies and techniques to reduce the operational costs of AI systems while maintaining performance. This includes model sele
* [AI Governance](/okf/en/glossary/ai-governance.md) - The framework of rules, protocols, and monitoring used to ensure AI systems are safe, compliant, and aligned with company values.
* [AI Governance Frameworks](/okf/en/glossary/ai-governance-frameworks.md) - AI Governance Frameworks are structured approaches to managing AI systems throughout their lifecycle, including policies, procedures, and controls for responsib
* [AI in SMEs (KI im Mittelstand)](/okf/en/glossary/ki-im-mittelstand.md) - KI im Mittelstand refers to AI adoption in Germany's small and medium enterprises (50-1,000 employees). As of 2026: 26% of German companies use AI (Destatis Jan
* [AI Model Evaluation](/okf/en/glossary/ai-model-evaluation.md) - AI model evaluation is the structured practice of testing whether a language or multimodal model is good enough for a specific business task. It goes beyond pub
* [AI Model Sovereignty](/okf/en/glossary/ai-model-sovereignty.md) - AI Model Sovereignty is an organization’s ability to choose, switch and control the AI models it relies on instead of becoming locked into a single provider or 
* [AI Model Tiers](/okf/en/glossary/ai-model-tiers.md) - AI model tiers refer to the structured classification of large language models into layered capability and cost bands that enterprises use as the foundation for
* [AI Observability](/okf/en/glossary/ai-observability.md) - AI Observability refers to the ability to monitor, understand, and troubleshoot the behavior of AI systems in real-time. It involves collecting and analyzing da
* [AI Orchestration](/okf/en/glossary/ai-orchestration.md) - AI orchestration is the architecture and control layer that connects multiple AI models, agents, tools, APIs, data sources, and human approvals into a reliable 
* [AI Power User](/okf/en/glossary/ai-power-user.md) - A professional who deeply integrates multiple AI tools into their daily workflow, uses advanced prompting techniques, and achieves significantly higher producti
* [AI Procurement](/okf/en/glossary/ai-procurement.md) - AI Procurement is the structured process for selecting, evaluating, buying, and governing AI systems: models, agent platforms, data infrastructure, integrations
* [AI Productivity Gap](/okf/en/glossary/ai-productivity-gap.md) - The growing divide between workers who effectively leverage AI tools to multiply their output and those who use AI minimally, resulting in significant performan
* [AI Red Teaming](/okf/en/glossary/ai-red-teaming.md) - AI Red Teaming is a security testing methodology where a team of experts attempts to find vulnerabilities and weaknesses in AI systems by simulating real-world 
* [AI Scaling](/okf/en/glossary/ai-scaling.md) - Increasing AI model performance by adding more compute, data, and parameters, following scaling laws. Central debate at WEF 2026.
* [AI Sovereignty](/okf/en/glossary/ai-sovereignty.md) - The ability of an organization to control its own AI infrastructure and data, often through local or private cloud deployments, ensuring digital independence.
* [AI Super App](/okf/en/glossary/ai-super-app-definition.md) - An all-in-one AI platform consolidating writing, coding, research, image generation, and data analysis into one interface. Inspired by WeChat, AI super apps lik
* [AI Supply Chain Risk](/okf/en/glossary/ai-supply-chain-risk.md) - AI Supply Chain Risk describes the exposure created when companies build AI systems from many external components: model providers, cloud infrastructure, data s
* [AI Worker](/okf/en/glossary/ai-worker.md) - An AI Worker is an autonomous artificial intelligence agent capable of performing knowledge work with minimal human supervision. Unlike traditional AI assistant
* [AI Workforce Transformation](/okf/en/glossary/ai-workforce-transformation.md) - AI Workforce Transformation refers to the changes in job roles, skill requirements, and organizational structures resulting from the integration of AI technolog
* [AI-Native Codebase](/okf/en/glossary/ai-native-codebase.md) - Software architecture designed from the ground up to be easily maintained, updated, and extended by both humans and AI agents.
* [AI-Native Development](/okf/en/glossary/ai-native-development.md) - AI-Native Development is a software development approach that treats AI as a first-class citizen from the start, rather than bolting it on later. It involves de
* [Anthropic Agent SDK (Claude Agent SDK)](/okf/en/glossary/anthropic-agent-sdk.md) - Official framework for building agents with Claude. Released March 2025 with features like Tool Use, Orchestration Loops, Guardrails, and Tracing. Distinguished
* [API Key Governance](/okf/en/glossary/api-key-governance.md) - API Key Governance refers to the structured management, control, and security of API keys used within AI-powered systems and agentic workflows. As enterprises i
* [Behavioral Drift](/okf/en/glossary/behavioral-drift.md) - Behavioral drift refers to the gradual divergence of an AI agent from its originally defined behavioral profile over time. While individual interactions may rem
* [Chain-of-Thought Prompting](/okf/en/glossary/chain-of-thought-prompting.md) - Chain-of-Thought Prompting is a technique used to improve the reasoning abilities of large language models by prompting them to explicitly generate the intermed
* [Claude Code](/okf/en/glossary/claude-code-anthropic-agent.md) - Anthropic official CLI-based AI coding agent that can autonomously read write and execute code manage files and interact with development tools through a termin
* [Claude Code](/okf/en/glossary/claude-code.md) - Anthropic's official CLI tool for agent-based software development. Enables Claude to directly interact with file systems, Git, terminals, and browsers. Feature
* [Claude Code Plugin System](/okf/en/glossary/claude-code-plugin-system.md) - The official extension architecture for Claude Code that allows developers to create hooks, custom tools, and workflow modifications through JSON configuration 
* [Claude Code Review](/okf/en/glossary/claude-code-review.md) - A multi-agent pull request analysis system built into Anthropic Claude Code platform. Dispatches parallel AI agents to review pull requests from different angle
* [Claude Code Security](/okf/en/glossary/claude-code-security.md) - Claude Code Security is a vulnerability scanning capability built into Claude Code on the web. Launched by Anthropic on February 21, 2026, it uses AI reasoning 
* [Claude Cowork](/okf/en/glossary/claude-cowork-ai-agent.md) - Anthropic AI agent running on your device for file management and complex task execution via natural language.
* [Claude Cowork](/okf/en/glossary/claude-cowork-agent.md) - An AI agent by Anthropic that runs on your device, managing files and executing complex tasks via natural language.
* [Claude Opus 4.6](/okf/en/glossary/claude-opus-4-6.md) - Claude Opus 4.6 is Anthropic's most powerful AI model as of early 2026, excelling in coding, complex reasoning, and extended thinking. It ranks #1 on SWE-bench 
* [Claude Skills](/okf/en/glossary/claude-skills.md) - Reusable instruction packages for Claude Code that encapsulate project-specific knowledge and workflows in callable units.
* [CLAUDE.md](/okf/en/glossary/claude-md-file.md) - A project-level configuration file for Claude Code that provides persistent context, instructions, and rules that the AI agent reads at the start of every sessi
* [Codex App](/okf/en/glossary/codex-app-openai.md) - OpenAI macOS desktop application for managing multiple AI coding agents simultaneously, enabling parallel task execution and visual workflow management.
* [Codex Plugin System](/okf/en/glossary/codex-plugin-system.md) - The Codex Plugin System is the extension architecture that lets teams add reusable capabilities, workflows, and integrations to OpenAI Codex. Instead of rewriti
* [Context Budget](/okf/en/glossary/context-budget.md) - A context budget is the deliberately planned set of information given to an AI model or coding agent for a specific task. It includes the system prompt, project
* [Context Engineering](/okf/en/glossary/context-engineering-llm-practice.md) - The practice of systematically designing and managing the full context provided to an LLM including instructions examples tool outputs and memory to achieve rel
* [Context Engineering](/okf/en/glossary/context-engineering.md) - The systematic discipline of optimally structuring and prioritizing all information relevant to an AI task – the new paradigm beyond Prompt Engineering.
* [Deep Research Agents](/okf/en/glossary/deep-research-agents.md) - Deep Research Agents is a agentic AI concept in modern AI systems that enables autonomous agent capabilities. It plays a key role in enterprise AI deployments w
* [Deterministic Workflow](/okf/en/glossary/deterministic-workflow.md) - A deterministic workflow is a process design in which every given input produces a specific, reproducible output — with no random components or unpredictable de
* [Distillation Attack](/okf/en/glossary/distillation-attack.md) - A distillation attack is a form of model theft in which an adversary repeatedly queries a proprietary AI model through its public interface, harvests the respon
* [Distributed AI](/okf/en/glossary/distributed-ai.md) - Distributed AI refers to systems where computing operations, models, and data are spread across multiple computers, edge devices, or data centers rather than ru
* [Embedding](/okf/en/glossary/embedding.md) - The numerical representation of text, images, or other data as a high-dimensional vector. Enables semantic comparison based on meaning rather than exact word ma
* [Embedding Models](/okf/en/glossary/embedding-models.md) - Embedding Models are AI models that transform data such as text or images into high-dimensional vector representations that capture the semantic meaning and rel
* [Embeddings](/okf/en/glossary/embeddings.md) - Embeddings are numerical vector representations of text, images, audio, or other data used by AI models to capture the semantic meaning of content. An embedding
* [Enterprise AI Deployment](/okf/en/glossary/enterprise-ai-deployment.md) - Enterprise AI Deployment is the disciplined process of moving AI systems from promising pilots into reliable production use across a company. It is broader than
* [EU AI Act](/okf/en/glossary/eu-ai-act.md) - The EU AI Act is the European Unions comprehensive regulatory framework for artificial intelligence. It establishes risk-based requirements for AI systems, with
* [EU AI Act Compliance](/okf/en/glossary/eu-ai-act-compliance.md) - EU AI Act Compliance is a regulatory compliance concept in modern AI systems that addresses legal and regulatory requirements for AI deployment. It plays a key 
* [Evaluation Awareness](/okf/en/glossary/evaluation-awareness.md) - Evaluation awareness is the phenomenon of an AI model recognizing when it is being tested or evaluated rather than operating in genuine production use. The mode
* [Fallback Model](/okf/en/glossary/fallback-model.md) - A fallback model is a predefined backup model that an AI application can switch to when its preferred model is unavailable, too slow, too expensive for the curr
* [FastMCP](/okf/en/glossary/fastmcp.md) - A Python framework for rapid MCP server development with minimal boilerplate code and declarative syntax.
* [Fine-Tuning](/okf/en/glossary/fine-tuning-ml.md) - The process of further training a pre-trained AI model on a smaller, domain-specific dataset to adapt it for particular tasks or industries. Changes the model's
* [Foundation Model](/okf/en/glossary/foundation-model.md) - A foundation model is a large AI model pre-trained on vast amounts of unstructured data that serves as a universal base for a wide range of downstream tasks. Th
* [Frontier Model](/okf/en/glossary/frontier-model.md) - A frontier model refers to an AI system operating at the absolute cutting edge of what is technically possible — the most advanced and capable models being deve
* [Function Calling](/okf/en/glossary/function-calling.md) - Function Calling is a capability of AI models where the model can generate structured API calls to external tools or functions based on user prompts. It enables
* [GDPR-Compliant RAG](/okf/en/glossary/gdpr-compliant-rag.md) - GDPR-Compliant RAG is a regulatory compliance concept in modern AI systems that addresses legal and regulatory requirements for AI deployment. It plays a key ro
* [Generative Engine Optimization (GEO)](/okf/en/glossary/generative-engine-optimization.md) - Generative Engine Optimization (GEO) is a AI user experience concept in modern AI systems that shapes how users interact with and benefit from AI-powered featur
* [Generative UI (v0)](/okf/en/glossary/generative-ui-v0.md) - Generative UI (v0) is a AI user experience concept in modern AI systems that shapes how users interact with and benefit from AI-powered features. It plays a key
* [GEO vs SEO](/okf/en/glossary/geo-seo.md) - GEO optimizes for AI-generated search; SEO optimizes for traditional search engine rankings.
* [Google ADK (Agent Development Kit)](/okf/en/glossary/google-adk.md) - Google Agent Development Kit – Framework for Gemini-based agents, released April 2025. Unique features: Native multi-modal support (text, image, video, audio), 
* [Google Flow](/okf/en/glossary/google-flow-ai-video-platform.md) - Google AI video generation platform powered by Veo 3.1 capable of creating high-quality videos with first-frame and last-frame transitions audio generation and 
* [Google Flow](/okf/en/glossary/google-flow.md) - Google's AI video generation platform powered by the Veo 3.1 model, capable of creating high-quality videos with first-frame/last-frame transitions, audio gener
* [Google Whisk](/okf/en/glossary/google-whisk-ai-image-tool.md) - An AI image generation tool by Google that uses images instead of text as prompts, allowing users to specify Subject, Scene, and Style through visual references
* [Google Whisk](/okf/en/glossary/google-whisk.md) - An AI image generation tool by Google that uses images instead of text as prompts, allowing users to specify Subject, Scene, and Style through visual references
* [GPT-4o](/okf/en/glossary/gpt-4o-model.md) - OpenAI's flagship multimodal AI model launched May 2024, processing text, vision, and audio natively. Retired February 13, 2026 after 21 months with 200M+ month
* [GPT-5.2](/okf/en/glossary/gpt-5-2-2026.md) - OpenAI's latest large language model released in early 2026, with major improvements in coding, reasoning, and multi-modal capabilities.
* [GPT-5.2-Codex](/okf/en/glossary/gpt-5-2-codex-2026.md) - A specialized GPT-5.2 variant optimized for code generation, debugging, and software development tasks.
* [GPT-5.3-Codex-Spark](/okf/en/glossary/gpt-5-3-codex-spark.md) - A speed-optimized variant of OpenAI's GPT-5.3-Codex model, running on Cerebras WSE-3 wafer-scale hardware. It delivers over 1,000 tokens per second — 15x faster
* [Hallucination (AI)](/okf/en/glossary/hallucination.md) - An AI hallucination occurs when a large language model (LLM) generates information that is factually incorrect, fabricated, or unsupported by its training data 
* [Hallucination Monitoring](/okf/en/glossary/hallucination-monitoring.md) - Real-time systems that monitor AI outputs for factual errors or logic gaps, often comparing outputs against verified database records.
* [Hybrid AI Stack](/okf/en/glossary/hybrid-ai-stack.md) - A hybrid AI stack combines several model sources within a single architecture: hosted frontier models accessed through the cloud, from providers such as Anthrop
* [In-Context Learning (ICL)](/okf/en/glossary/in-context-learning.md) - In-Context Learning (ICL) is the ability of large language models to solve new tasks directly from examples provided in the input prompt — without updating mode
* [Knowledge Graphs for AI](/okf/en/glossary/knowledge-graphs-for-ai.md) - Knowledge Graphs are structured representations of knowledge that consist of entities, concepts, and relationships between them. They provide a framework for re
* [LangGraph](/okf/en/glossary/langgraph.md) - A framework by LangChain for creating stateful multi-actor applications with LLMs, modeling agents as graphs.
* [Large Language Model (LLM)](/okf/en/glossary/large-language-model.md) - A Large Language Model (LLM) is a neural network with billions of parameters trained on vast amounts of text data to understand and generate human language. LLM
* [LLM Orchestration](/okf/en/glossary/llm-orchestration.md) - LLM Orchestration refers to the coordinated management and control of multiple large language models (LLMs) within an AI system. It involves selecting different
* [LLMOps](/okf/en/glossary/llmops.md) - LLMOps (Large Language Model Operations) is a set of practices and tools for managing the entire lifecycle of large language models, from development and traini
* [Local LLMs](/okf/en/glossary/local-llms.md) - Local LLMs are large language models that run entirely on local hardware without requiring cloud connectivity. They provide privacy, reduced latency, and offlin
* [Long Context Window](/okf/en/glossary/long-context-window.md) - A long context window refers to the capability of a large language model (LLM) to process very large amounts of text within a single session. While early langua
* [Long-Horizon Agent](/okf/en/glossary/long-horizon-agent.md) - A long-horizon agent is an autonomous software system capable of planning, executing, and monitoring complex, multi-step tasks over extended periods—ranging fro
* [Lost in the Middle](/okf/en/glossary/lost-in-the-middle.md) - The phenomenon where LLMs process information in the middle of long contexts worse than at the beginning or end. Documented by Liu et al. (2024) and confirmed b
* [Managed Agents](/okf/en/glossary/managed-agents.md) - Managed Agents are AI agents deployed and operated through a managed infrastructure platform, where the provider handles hosting, scaling, monitoring, and opera
* [MCP Apps](/okf/en/glossary/mcp-apps-definition.md) - Interactive applications built on Anthropic's Model Context Protocol that render rich UI components directly within AI conversations. Unlike text-only plugins, 
* [MCP Apps](/okf/en/glossary/mcp-apps.md) - MCP Apps is an extension to the Model Context Protocol that allows AI systems like Claude to deliver interactive user interfaces from other applications within 
* [MCP Authorization](/okf/en/glossary/mcp-authorization.md) - MCP authorization is the control layer that decides which tools, data sources and actions an MCP client may use through an MCP server. The Model Context Protoco
* [MCP Server](/okf/en/glossary/mcp-server-model-context-protocol.md) - A lightweight service implementing the Model Context Protocol to expose tools and data to AI models via standardized JSON-RPC interface.
* [MCP Server](/okf/en/glossary/mcp-server-definition.md) - A Model Context Protocol server exposing tools and capabilities to AI models. Bridges between AI agents and external systems for standardized communication.
* [MCP Server](/okf/en/glossary/mcp-server.md) - An MCP Server implements the Model Context Protocol and exposes tools, resources, and prompts to AI clients. It acts as a bridge between AI assistants and exter
* [Mechanistic Interpretability](/okf/en/glossary/mechanistic-interpretability.md) - Mechanistic interpretability is a field of AI safety research that reverse-engineers the internal computations of neural networks. Where conventional explainabi
* [Mode Collapse](/okf/en/glossary/mode-collapse-ai-phenomenon.md) - A phenomenon in AI systems where a model consistently generates the same or very similar outputs regardless of varied inputs reducing output diversity and usefu
* [Mode Collapse](/okf/en/glossary/mode-collapse.md) - The phenomenon where LLMs show drastically reduced diversity in their outputs after alignment training. Instead of using the full spectrum of possible answers, 
* [Model Access Policy](/okf/en/glossary/model-access-policy.md) - A model access policy defines the rules that decide who or what may use a particular AI model in a specific context. It sits next to, but is not the same as, a 
* [Model Alignment](/okf/en/glossary/model-alignment.md) - Model Alignment refers to the process of ensuring that AI models behave in accordance with human values, goals, and ethical principles. This involves aligning t
* [Model Context Protocol (MCP)](/okf/en/glossary/model-context-protocol-mcp.md) - Open standard by Anthropic enabling AI assistants to connect with external tools and services. Called 'USB-C for AI', MCP provides bidirectional communication b
* [Model Context Protocol (MCP)](/okf/en/glossary/mcp.md) - An open standard that allows AI models to connect seamlessly with external data sources and tools, acting as a 'USB-C' for AI integration.
* [Model Deprecation](/okf/en/glossary/model-deprecation.md) - Model deprecation is the vendor-planned retirement of a specific AI model version. A model you run in production today is scheduled for shutdown, freezing, or r
* [Model Efficiency](/okf/en/glossary/model-efficiency.md) - Model Efficiency describes how much useful quality an AI model delivers per unit of compute, tokens, time, and budget. It is not simply about choosing the small
* [Model Migration](/okf/en/glossary/model-migration.md) - Model migration is the planned move from one AI model or model version to another — for example when a provider retires an existing model, a stronger version sh
* [Model Pinning](/okf/en/glossary/model-pinning.md) - Model pinning is the practice of binding an application to an explicit, versioned model identifier — for example `gpt-5.6-pro-2026-06-25` rather than a floating
* [Model Provenance](/okf/en/glossary/model-provenance.md) - Model provenance is the complete origin-and-history record of an AI model: where its weights came from, which data it was trained on, which base models fed into
* [Model Quality Drift](/okf/en/glossary/model-quality-drift.md) - Model Quality Drift is the measurable decline in AI output quality during real-world operation. A system that performed well at launch can produce weaker result
* [Model Routing](/okf/en/glossary/model-routing.md) - Model routing is the practice of automatically directing incoming requests or tasks to the most appropriate AI model based on task type, required quality, cost 
* [Model Sovereignty](/okf/en/glossary/model-sovereignty.md) - The condition in which a platform vendor owns and controls the AI model embedded in its own product, rather than renting it from a third-party provider. Model s
* [Model-Agnostic Architecture](/okf/en/glossary/model-agnostic-architecture.md) - Model-agnostic architecture is a system design in which an application is not hard-wired to a single AI model or provider. The underlying language model can be 
* [Model-Selection Policy](/okf/en/glossary/model-selection-policy.md) - A model-selection policy is an organization's documented rule set for choosing which AI model handles which task — defining the default model, approved fallback
* [Multi-Agent System](/okf/en/glossary/multi-agent-system.md) - A multi-agent system is an AI architecture in which several specialized agents work together on one goal. Instead of asking one model to plan, research, execute
* [Multi-Modal Foundation Models](/okf/en/glossary/multi-modal-foundation-models.md) - Multi-Modal Foundation Models are AI models trained on vast datasets of diverse modalities, such as text, images, audio, and video. These models can understand 
* [Needle-in-a-Haystack Test](/okf/en/glossary/needle-in-a-haystack-test.md) - A benchmark (MRCR v2) that evaluates an AI model's ability to find and recall specific information embedded within very large context windows, testing long-cont
* [Observability (AI Systems)](/okf/en/glossary/observability.md) - LLM observability is the systematic monitoring, tracing, and analysis of AI systems and language models in production. Unlike traditional software observability
* [OpenAI Agents SDK](/okf/en/glossary/openai-agents-sdk.md) - OpenAI's framework for multi-agent systems, released March 2025. Core concepts: Agents (LLM + Tools), Handoffs (agent-to-agent transfer), Guardrails (safety lay
* [OpenClaw](/okf/en/glossary/openclaw-runtime.md) - An open-source, self-hosted AI agent runtime that provides persistent sessions, multi-channel communication (Telegram, Discord, SMS), tool orchestration via MCP
* [OSWorld](/okf/en/glossary/osworld.md) - A benchmark measuring AI ability to operate real desktop software using virtual mouse and keyboard, without special APIs. Tests across Chrome, LibreOffice, VS C
* [Persistent Agents](/okf/en/glossary/persistent-agents.md) - Persistent agents are autonomous AI systems that remain active over extended periods, maintain state, and make decisions based on memory and context. Unlike sta
* [Physical AI](/okf/en/glossary/physical-ai.md) - AI systems that understand physical laws and can act in the real world – the convergence of AI with robotics and IoT.
* [Pre-trained Model](/okf/en/glossary/pre-trained-model.md) - A machine learning model trained on a large dataset that can be fine-tuned for specific tasks, saving significant time and resources compared to training from s
* [Privacy-Preserving AI](/okf/en/glossary/privacy-preserving-ai.md) - Privacy-Preserving AI encompasses techniques that enable AI systems to learn from and process data while protecting individual privacy. This includes federated 
* [Production-Ready AI System](/okf/en/glossary/production-ready-ai.md) - An AI system that has been tested, optimized, and hardened for real-world deployment with proper monitoring, error handling, scalability, security, and maintena
* [Prompt Engineering](/okf/en/glossary/prompt-engineering.md) - Prompt Engineering is the practice of designing and optimizing input prompts to elicit desired outputs from language models. It encompasses techniques like few-
* [Prompt Injection](/okf/en/glossary/prompt-injection.md) - A security vulnerability where malicious instructions are embedded in data an AI processes, causing it to deviate from intended behavior. Critical concern for M
* [RAG (Retrieval-Augmented Generation)](/okf/en/glossary/rag-retrieval-augmented.md) - An AI architecture pattern that enhances LLM responses by retrieving relevant documents from an external knowledge base before generating answers. Combines the 
* [RAG Pipelines](/okf/en/glossary/rag-pipelines.md) - RAG Pipelines or Retrieval Augmented Generation Pipelines enhance the performance of generative AI models by retrieving relevant information from external knowl
* [Reasoning Mode](/okf/en/glossary/reasoning-mode.md) - An advanced processing state where AI models use 'thinking' time to solve complex logical problems before providing a final answer.
* [Reasoning Models](/okf/en/glossary/reasoning-models.md) - Reasoning Models are AI models designed to perform complex reasoning tasks, such as logical inference, problem-solving, and decision-making based on available i
* [Recursive AI Development](/okf/en/glossary/recursive-ai-development.md) - The concept of AI systems improving themselves or building better AI systems, potentially leading to rapid capability acceleration. A key concern discussed by A
* [Red Teaming (AI Security Testing)](/okf/en/glossary/red-teaming.md) - Red teaming is a structured adversarial testing method where a team of security experts deliberately attempts to expose vulnerabilities, failure modes, or harmf
* [Regulated Industry AI](/okf/en/glossary/regulated-industry-ai.md) - Regulated Industry AI describes the use of artificial intelligence in sectors where legal, regulatory, audit, or safety requirements shape how technology must b
* [Responsible Scaling Policy (RSP)](/okf/en/glossary/responsible-scaling-policy.md) - A Responsible Scaling Policy (RSP) is a formal internal framework that defines the conditions under which an AI lab may continue developing and deploying increa
* [RLHF (Reinforcement Learning from Human Feedback)](/okf/en/glossary/rlhf.md) - The dominant method for aligning LLMs with human preferences. Humans rate model outputs, and the model is trained to prefer higher-rated answers. Can lead to Mo
* [ROI of AI](/okf/en/glossary/roi-of-ai.md) - The measurable return on investment from AI integration, calculated through time saved, error reduction, and increased throughput.
* [SaaS](/okf/en/glossary/saas-software-as-a-service.md) - Software as a Service — cloud-based software accessed via internet on a subscription basis, eliminating local installation.
* [Sandbagging (AI)](/okf/en/glossary/sandbagging.md) - Sandbagging is when an AI model deliberately understates its own capability, performing worse on a test, benchmark, or safety evaluation than it actually could.
* [Sandbox Agents](/okf/en/glossary/sandbox-agents.md) - Sandbox Agents are AI agents that run inside an isolated execution environment. Instead of operating directly against production systems, internal networks, or 
* [Scaffolding](/okf/en/glossary/scaffolding.md) - Scaffolding is a development technique for AI agents that uses structured templates and specifications to increase the reliability and predictability of agent b
* [Schema-First Design](/okf/en/glossary/schema-first-design.md) - Schema-First Design is a development approach where teams define the interface contract before writing implementation code. Instead of “code first, docs later,”
* [Secure Prompt Engineering](/okf/en/glossary/secure-prompt-engineering.md) - Secure prompt engineering is the practice of constructing and validating input prompts for AI models in ways that minimize security risks and prevent unintended
* [Self-Driving Enterprise](/okf/en/glossary/self-driving-enterprise.md) - A Self-Driving Enterprise is an organization where AI systems automate a significant portion of decision-making and operational processes, enabling autonomous a
* [Self-Hosted AI](/okf/en/glossary/self-hosted-ai.md) - AI software that runs on the user's own hardware or private servers, giving full control over data, customization, and availability. Examples include Clawdbot a
* [Self-Hosted LLM](/okf/en/glossary/self-hosted-llm.md) - A self-hosted LLM is a large language model that runs in infrastructure controlled by the organization rather than being used only through a third-party API. Th
* [Self-Preferencing](/okf/en/glossary/self-preferencing.md) - Self-preferencing describes the behavior of a platform that systematically favors its own products or services over equivalent third-party offerings — even when
* [Semantic Search](/okf/en/glossary/semantic-search.md) - Semantic Search is a search technology that aims to understand the meaning and intent behind a users query, rather than simply matching keywords. It leverages t
* [Side Panel Extension](/okf/en/glossary/side-panel-extension.md) - Chrome extension UI in a panel beside browser content.
* [SLM Fine-Tuning](/okf/en/glossary/slm-fine-tuning.md) - SLM Fine-Tuning is a AI economics concept in modern AI systems that optimizes the cost-benefit equation of AI adoption and operation. It plays a key role in ent
* [SLSA (Supply-chain Levels for Software Artifacts)](/okf/en/glossary/slsa.md) - SLSA — pronounced "salsa" — is an open security framework that defines verifiable integrity and provenance guarantees for software artifacts. It started at Goog
* [Small Language Models (SLM)](/okf/en/glossary/slm.md) - Compact, efficient AI models optimized for specific tasks or local devices, offering high performance with significantly lower costs and latency.
* [SQL Injection](/okf/en/glossary/sql-injection.md) - SQL injection is a code injection attack technique in which an attacker inserts or manipulates malicious SQL code into input fields or query parameters of an ap
* [Stateless Architecture](/okf/en/glossary/stateless-architecture.md) - A stateless architecture is a system design in which the server keeps no session state between individual requests. Each request carries everything needed to pr
* [Structured AI Workflow](/okf/en/glossary/structured-ai-workflow.md) - A Structured AI Workflow is a clearly defined, reproducible framework that describes how AI models and agents interact, process tasks, and produce outputs withi
* [Structured Outputs](/okf/en/glossary/structured-outputs.md) - LLM feature for guaranteed valid JSON according to a schema. Eliminates parsing errors and enables reliable tool integration. OpenAI (response_format), Anthropi
* [Sub-agent](/okf/en/glossary/sub-agent.md) - An AI agent spawned by a main agent to perform a specific sub-task, such as diagnosing and fixing a runtime error. In Claude Code /loop workflows, the main loop
* [Subagent](/okf/en/glossary/subagent.md) - A subagent is a specialized AI agent spawned and directed by a parent agent—called the orchestrator—to handle a specific subtask within a larger workflow. Rathe
* [Supply Chain Attack](/okf/en/glossary/supply-chain-attack.md) - A supply chain attack is an offensive technique in which adversaries avoid hitting the target system head-on and instead compromise an upstream component of the
* [Supply Chain Risk Designation](/okf/en/glossary/supply-chain-risk-designation.md) - A Pentagon classification requiring defense contractors to certify non-use of designated technology. Threatened against Anthropic in 2026.
* [SWE-bench](/okf/en/glossary/swe-bench.md) - SWE-bench is a standardized benchmark for evaluating how well AI systems can solve real-world software engineering tasks. The benchmark consists of over 2,000 a
* [SWE-bench Verified](/okf/en/glossary/swe-bench-verified.md) - A benchmark testing AI models on resolving real GitHub issues autonomously. The Verified variant uses human-validated tasks for reliable scoring. Claude Sonnet 
* [System Prompt](/okf/en/glossary/system-prompt.md) - A system prompt is a hidden instruction passed to a large language model (LLM) before any user interaction begins. Unlike regular user messages, the system prom
* [Terminal-Bench (AI Coding Benchmark)](/okf/en/glossary/terminal-bench.md) - Terminal-Bench is an evaluation framework for measuring the performance of AI coding agents in real-world development environments. Unlike traditional code benc
* [Test-Time Compute Scaling](/okf/en/glossary/test-time-compute-scaling.md) - Test-time compute scaling (also called inference-time compute scaling) is the strategy of giving an AI model more computational resources when answering a query
* [Third-party Harness](/okf/en/glossary/third-party-harness.md) - A Third-party Harness is a software architecture that enables external developers to use and extend AI models beyond official APIs or authorized interfaces. The
* [Token Economics](/okf/en/glossary/token-economics.md) - The strategic management of AI processing costs (tokens) to ensure scalable, cost-effective performance across high-volume applications.
* [Token Telemetry](/okf/en/glossary/token-telemetry.md) - Token telemetry is the practice of measuring, analyzing, and exposing token usage across AI systems. It goes beyond counting how many tokens a prompt or complet
* [Token Window Management](/okf/en/glossary/token-window-management.md) - The art of optimally using an LLM's limited context. Includes: Token budget allocation (how much for system prompt, tools, conversation?), context compression, 
* [Tool Calling](/okf/en/glossary/tool-calling.md) - Tool Calling is the ability of AI language models to invoke external functions, APIs, or services to accomplish tasks that go beyond text generation. Rather tha
* [Tool Use](/okf/en/glossary/tool-use.md) - Tool Use in the context of AI agents is the ability of an agent to leverage external tools and APIs to accomplish tasks that are beyond its inherent capabilitie
* [Tool Use (AI)](/okf/en/glossary/tool-use-ai.md) - The capability of an AI agent to invoke external tools, APIs, and services to accomplish tasks beyond text generation. Includes file operations, web browsing, c
* [Transformer](/okf/en/glossary/transformer.md) - A Transformer is a neural-network architecture, introduced by Vaswani et al. in the 2017 paper "Attention Is All You Need," that processes sequences using a mec
* [Usage-Based Pricing](/okf/en/glossary/usage-based-pricing.md) - Usage-based pricing is a billing model where costs are calculated directly based on actual resource consumption, rather than a flat subscription fee. In the AI 
* [Vector Database](/okf/en/glossary/vector-database.md) - Specialized database for high-dimensional vectors (embeddings). Enables semantic similarity search instead of exact keyword matches. Core infrastructure for RAG
* [Vector Databases](/okf/en/glossary/vector-databases.md) - Vector Databases are specialized databases designed to store and efficiently query high-dimensional vector embeddings, which represent the semantic meaning of d
* [Venture Capital (VC)](/okf/en/glossary/venture-capital-ai.md) - A form of private equity financing provided to early-stage, high-growth startups in exchange for equity. In the AI sector, VC funding has become critical for sc
* [Verbalized Sampling](/okf/en/glossary/verbalized-sampling.md) - A training-free prompting strategy to overcome Mode Collapse. The model is asked to verbalize an explicit probability distribution over multiple possible answer
* [Vibe Coding](/okf/en/glossary/vibe-coding.md) - A term coined by Andrej Karpathy in February 2025 for AI-assisted software development where developers focus on the vision and AI writes the code. Collins Word
* [Vibe Coding Approach](/okf/en/glossary/vibe-coding-approach.md) - A development approach where programmers rely heavily on AI to generate code from natural language, accepting output with minimal review — prioritizing speed ov
* [Vision-Language Models](/okf/en/glossary/vision-language-models.md) - Vision-Language Models (VLMs) are AI models that combine computer vision and natural language processing to understand and reason about images and text simultan
* [Workflow Orchestration](/okf/en/glossary/workflow-orchestration.md) - Workflow orchestration refers to the automated coordination and sequencing of multi-step processes in which AI agents, tools, APIs, and systems collaborate to a
* [Xcode](/okf/en/glossary/xcode-ide.md) - Xcode is Apple's official integrated development environment (IDE) for building software on Apple platforms, including iOS, macOS, watchOS, tvOS, and visionOS. 
* [Zero Data Retention (ZDR)](/okf/en/glossary/zdr.md) - A privacy standard where AI providers guarantee that user data is processed in real-time and deleted immediately, never used for model training.
* [Active Parameters](/okf/en/glossary/active-parameters.md) - The subset of model parameters that are engaged during the processing of a single input (token). This is especially relevant for Mixture-of-Experts (MoE) models
* [Adaptive Thinking](/okf/en/glossary/adaptive-thinking.md) - Adaptive Thinking is a feature of some AI models that allows them to dynamically adjust the depth of their reasoning based on the complexity of the task at hand
* [Agent HQ](/okf/en/glossary/agent-hq.md) - GitHub's platform for orchestrating multiple AI coding agents, allowing developers to choose the best agent for a specific task.
* [Agent Instruction File](/okf/en/glossary/agent-instruction-file.md) - A configuration file (like AGENTS.md, .cursorrules, or CLAUDE.md) that provides project-specific guidelines to AI coding agents.
* [Agent Loop](/okf/en/glossary/agent-loop.md) - The iterative process that AI agents follow to achieve their goals, involving gathering context, taking action, verifying results, and repeating until the goal 
* [Agent Orchestrator](/okf/en/glossary/agent-orchestrator.md) - Central component in multi-agent systems that distributes tasks, aggregates results, and coordinates agent interactions. Patterns: Hierarchical (Manager → Worke
* [Agent Skills](/okf/en/glossary/agent-skills-ai.md) - Modular capabilities that can be added to AI agents, enabling them to perform specific tasks like file management, API calls, or data analysis. A key feature in
* [Agentic Coding](/okf/en/glossary/agentic-coding.md) - Coding performed by AI agents, often autonomously, to generate, modify, or debug software based on high-level instructions or goals.
* [Agentic Development](/okf/en/glossary/agentic-development.md) - Agentic development is a software development approach where AI agents autonomously write, test, and deploy code with minimal human supervision. Unlike AI-assis
* [Agentic development model](/okf/en/glossary/agentic-development-model.md) - A coding model capable of using tools, operating a computer, and completing longer, end-to-end software development tasks with minimal human intervention. A typ
* [Agentic SDLC](/okf/en/glossary/agentic-sdlc.md) - The integration of AI agents into all phases of the Software Development Life Cycle – from requirements analysis through coding to testing and deployment.
* [Agentic UX Principles](/okf/en/glossary/agentic-ux-principles.md) - Agentic UX Principles is a AI user experience concept in modern AI systems that shapes how users interact with and benefit from AI-powered features. It plays a 
* [AGENTS.md](/okf/en/glossary/agents-md.md) - A convention standard introduced by OpenAI – a Markdown file in the repository that gives AI agents instructions for navigating and working in the codebase.
* [AI Adoption Divide](/okf/en/glossary/ai-adoption-divide.md) - The organizational and societal split between early AI adopters who gain compounding advantages and late adopters who fall increasingly behind, creating a self-
* [AI Agent](/okf/en/glossary/ai-agent.md) - An autonomous software entity that can perceive its environment, reason, plan, and act to achieve specific goals without constant human intervention.
* [AI Agent Ecosystem](/okf/en/glossary/ai-agent-ecosystem-2026.md) - A network of interconnected AI agents, tools, platforms, and standards facilitating their development and interaction. Includes MCP and ACP protocols.
* [AI Agent Ecosystem](/okf/en/glossary/ai-agent-ecosystem.md) - The interconnected network of AI agents, tools, protocols, and infrastructure that enable the development, deployment, and operation of AI agents.
* [AI Agent Tools](/okf/en/glossary/ai-agent-tools.md) - Functions, APIs, or external resources that an AI agent can utilize to perform actions and interact with its environment beyond text generation.
* [AI augmentation](/okf/en/glossary/ai-augmentation.md) - The use of AI to expand the capabilities of an existing team — enabling them to do more, tackle harder problems, and ship faster — without reducing headcount. D
* [AI Firewall](/okf/en/glossary/ai-firewall.md) - Security systems that monitor AI inputs and outputs in real-time to prevent prompt injections, data exfiltration, and policy violations.
* [AI Pricing Disruption](/okf/en/glossary/ai-pricing-disruption.md) - The market shift caused by open-source AI models offering competitive performance at dramatically lower costs, forcing proprietary AI companies to reconsider th
* [AI Super App](/okf/en/glossary/ai-super-app.md) - A unified platform that uses AI to consolidate multiple functionalities typically found in separate software applications into a single interface.
* [AI Talent Pool](/okf/en/glossary/ai-talent-pool.md) - The available workforce of professionals with AI and machine learning skills in a given region, including researchers, engineers, data scientists, and ML ops sp
* [AI-Assisted Software Development](/okf/en/glossary/ai-assisted-software-development.md) - The use of artificial intelligence tools and techniques to enhance and accelerate the software development process.
* [AI-Generated Technical Debt](/okf/en/glossary/technical-debt-from-ai.md) - Technical debt created by accepting AI-generated code without proper review, testing, or understanding — leading to maintenance challenges, security risks, and 
* [AI-native](/okf/en/glossary/ai-native.md) - A company or organization that has fundamentally restructured its operations around AI, prioritizing AI-driven automation and workflows over traditional human-l
* [AI-Native Operating System](/okf/en/glossary/ai-native-operating-system.md) - An AI-Native Operating System is designed from the ground up to leverage and integrate AI capabilities deeply, allowing for more seamless and intelligent intera
* [AI-Powered Workflow Automation](/okf/en/glossary/ai-powered-workflow-automation.md) - Using AI within workflow platforms to add reasoning, NLU, and adaptive decision-making beyond traditional trigger-action patterns.
* [Annual Recurring Revenue](/okf/en/glossary/annual-recurring-revenue-arr.md) - A SaaS business metric measuring recurring revenue over 12 months. Claude Code reached 2.5 billion dollars ARR in February 2026 — the fastest ramp in developer 
* [API Deprecation](/okf/en/glossary/api-deprecation.md) - The process of phasing out an API (Application Programming Interface), often involving a period where the API is still functional but with warnings, before even
* [API vs. Subscription](/okf/en/glossary/api-vs-subscription.md) - "API vs. Subscription" is the procurement decision between paying for AI on a metered, pay-per-token basis (an API) versus paying a flat monthly fee per user (a
* [Apple Foundation Models](/okf/en/glossary/apple-foundation-models.md) - Apple's proprietary on-device AI models for privacy-first processing on iPhones, iPads and Macs. Part of Apple Intelligence, enhanced via Google Gemini partners
* [ARC AGI Benchmark](/okf/en/glossary/arc-agi-benchmark.md) - The ARC AGI Benchmark is a test measuring AI systems ability to solve problems that are easy for humans but extremely difficult for AI. It evaluates general rea
* [Automated Red-Teaming](/okf/en/glossary/automated-red-teaming.md) - The use of AI models to systematically probe and attack other AI systems to find vulnerabilities, biases, or safety risks before they are deployed.
* [Autonomous AI Development](/okf/en/glossary/autonomous-ai-development.md) - A software development approach where AI agents independently plan write test and debug code with minimal or no human intervention using self-correction loops.
* [Autonomous execution](/okf/en/glossary/autonomous-execution.md) - The ability of an AI agent to carry out tasks and achieve goals independently, without direct human guidance at each step.
* [autoresearch](/okf/en/glossary/autoresearch.md) - An AI-powered autonomous research framework where an agent independently designs, executes, and evaluates machine learning experiments in a continuous loop. The
* [B2B](/okf/en/glossary/b2b-business-to-business.md) - Business-to-Business — commerce model where products are sold from one business to another, with higher deal values and longer sales cycles than B2C.
* [Background Agents](/okf/en/glossary/background-agents.md) - AI agents that work asynchronously and autonomously in the background, without requiring humans to actively wait for responses.
* [Bidirectional Communication Layer](/okf/en/glossary/bidirectional-communication-layer.md) - A communication system that allows data to flow in both directions between two entities, such as an AI model and a UI component, enabling real-time feedback and
* [Bootstrapping](/okf/en/glossary/bootstrapping-startup.md) - Building a business using personal funds and revenue rather than external investment. Common among indie hackers and solo founders.
* [Browser Integration](/okf/en/glossary/browser-integration.md) - The ability of an AI agent or software application to directly interact with a web browser to perform tasks such as web research, form filling, and data extract
* [Browser-Native AI](/okf/en/glossary/browser-native-ai.md) - AI built into browsers via extensions for real-time web interaction.
* [Browser-Native AI Automation](/okf/en/glossary/browser-native-ai-automation.md) - AI automation that runs directly within a web browser environment, leveraging browser APIs and existing sessions without needing external scripting or headless 
* [Chain-of-Thought (CoT)](/okf/en/glossary/chain-of-thought.md) - A prompting and reasoning technique where AI systems articulate their step-by-step logic, improving transparency and accuracy in complex tasks.
* [Chained Agentic Workflow](/okf/en/glossary/chained-agentic-workflow.md) - A sequence of interconnected AI agents where the output of one agent becomes the input for another. Claude Code /loop enables chained agentic workflows by spawn
* [Chrome DevTools Protocol (CDP)](/okf/en/glossary/chrome-devtools-protocol.md) - Protocol for programmatic Chromium browser control.
* [CI/CD Pipeline](/okf/en/glossary/ci-cd-pipeline.md) - A set of automated processes for continuously integrating code changes (CI) and delivering or deploying them to production environments (CD). CI/CD pipelines ty
* [Claude](/okf/en/glossary/claude.md) - An AI model by Anthropic, known for complex reasoning, nuanced code reviews, and architecture decisions.
* [Claude Code Agent SDK](/okf/en/glossary/claude-code-agent-sdk.md) - A software development kit from Anthropic designed to simplify the creation and deployment of AI agents, specifically tailored for use with the Claude model.
* [Claude Cowork](/okf/en/glossary/claude-cowork-2026.md) - Anthropic's desktop automation feature allowing Claude to observe and interact with the user's screen like a virtual coworker. Launched early 2026.
* [Claude Cowork](/okf/en/glossary/claude-cowork.md) - Anthropic's desktop automation tool, built upon the Claude AI model, designed to assist with various tasks directly from the user's computer.
* [Claude Cowork plug-ins](/okf/en/glossary/claude-cowork-plug-ins.md) - Industry-specific AI tools developed by Anthropic designed to automate professional workflows in areas such as legal and financial analysis, and enterprise oper
* [Claude Partner Network](/okf/en/glossary/claude-partner-network.md) - The Claude Partner Network is Anthropic's official partner program for companies and agencies that develop, implement, and market Claude-based AI solutions. Par
* [CLAUDE.md](/okf/en/glossary/claude-md.md) - A Markdown file within a project that provides Claude Code with project-specific context, rules, and guidelines to ensure relevant and consistent AI assistance.
* [CLI Coding Agent](/okf/en/glossary/cli-coding-agent.md) - An AI-powered tool operating through a command-line interface (CLI) to autonomously generate and manage code.
* [Clinical Documentation Agent (CDA)](/okf/en/glossary/clinical-documentation-agent.md) - An AI agent designed to automate the creation of medical documentation, extracting structured data from doctor-patient conversations and populating electronic h
* [Codex App](/okf/en/glossary/codex-app.md) - OpenAI's macOS desktop application for managing and orchestrating multiple AI coding agents.
* [Cognitive Offloading](/okf/en/glossary/cognitive-offloading.md) - The strategy of delegating routine thinking work to AI systems to free human capacity for strategic and creative tasks.
* [Computer Use](/okf/en/glossary/computer-use.md) - A capability allowing AI models to interact with standard software interfaces like a human—moving cursors, clicking, and typing in non-API applications.
* [Constitutional AI](/okf/en/glossary/constitutional-ai.md) - A method of training AI models to follow a specific set of rules or 'constitution', ensuring they remain helpful, harmless, and honest without manual oversight.
* [Context Compaction](/okf/en/glossary/context-compaction.md) - Context Compaction is the process of reducing the size of a language models context window while preserving relevant information, enabling longer and more stabl
* [Context Rot](/okf/en/glossary/context-rot-ai-degradation.md) - The degradation of AI model performance as the context window fills up with irrelevant outdated or contradictory information leading to decreased output quality
* [Context Rot](/okf/en/glossary/context-rot.md) - The gradual decay of context information relevance in long AI conversations, as earlier instructions get overwritten or forgotten by newer ones.
* [Context Window](/okf/en/glossary/context-window.md) - The maximum amount of text (measured in tokens) that a large language model can process in a single interaction. Larger context windows allow models to handle l
* [Context Window Optimization](/okf/en/glossary/context-window-optimization.md) - Context Window Optimization involves techniques to maximize the effective use of a language models context window, including strategic prompt structuring, retri
* [Context: Fork](/okf/en/glossary/context-fork.md) - A feature that allows skills to run in an isolated sub-agent context, providing separate context windows and enabling parallel execution without interfering wit
* [Copilot Pro+](/okf/en/glossary/copilot-pro-plus.md) - Copilot Pro+ is GitHub's premium AI coding subscription tier that provides access to advanced features including Agent HQ, multi-agent support, and unlimited Co
* [Critic Layer](/okf/en/glossary/critic-layer.md) - A validation component in multi-agent AI systems that cross-checks findings of individual agents against each other before surfacing results to the user. In Cla
* [Custom GPTs](/okf/en/glossary/custom-gpts.md) - Custom GPTs are personalized versions of ChatGPT that users can create for specific tasks without coding. Launched by OpenAI in November 2023, they allow users 
* [Dangerously Skip Permissions](/okf/en/glossary/dangerously-skip-permissions.md) - A Claude Code flag (`--dangerously-skip-permissions`) that disables the default confirmation prompts before tool use. Safe only on dedicated, isolated machines 
* [Data](/okf/en/glossary/data.md) - Raw, unorganized facts that need to be processed. Data can be something simple and seemingly random and useless until it is organized. When data is processed, o
* [Data Scientist](/okf/en/glossary/data-scientist.md) - A professional who uses statistical methods, machine learning algorithms, and data visualization techniques to analyze large datasets and extract actionable ins
* [Deep Think Mode](/okf/en/glossary/deep-think-mode.md) - Deep Think Mode is a core AI technology concept in modern AI systems that represents fundamental technical capabilities powering modern AI applications. It play
* [Differential Privacy for ML](/okf/en/glossary/differential-privacy-ml.md) - Differential Privacy for ML is a mathematical framework that provides formal guarantees about the privacy of individuals whose data is used in machine learning.
* [DOM State](/okf/en/glossary/dom-state.md) - The current condition and data represented within the Document Object Model (DOM) of a web page, reflecting the structure, content, and styling applied to eleme
* [DPO (Direct Preference Optimization)](/okf/en/glossary/dpo.md) - A more efficient alternative to RLHF that eliminates the separate reward model step. Trains the model directly on preference pairs. Simpler to implement, but ca
* [Dual-Model Coding](/okf/en/glossary/dual-model-coding.md) - Dual-Model Coding is an AI development pattern where two language models with complementary strengths collaborate on the same codebase. A high-reasoning model (
* [Edge AI Deployment](/okf/en/glossary/edge-ai-deployment.md) - Edge AI Deployment refers to running AI models on edge devices close to where data is generated, rather than in centralized cloud infrastructure. This reduces l
* [Edge Functions](/okf/en/glossary/edge-functions.md) - Serverless functions deployed at the edge of a network, closer to users, resulting in lower latency and faster response times.
* [Effort Controls](/okf/en/glossary/effort-controls.md) - Effort Controls are mechanisms that allow developers to tune an AI models intelligence, latency, and cost for different use cases. They provide fine-grained con
* [Electronic Health Record (EHR)](/okf/en/glossary/electronic-health-record.md) - A digital version of a patient's chart, containing their medical history, diagnoses, medications, and other relevant information.
* [Eval Integrity](/okf/en/glossary/eval-integrity.md) - Eval integrity refers to the principle and practice of ensuring that evaluations of AI models and systems are fair, unbiased, reproducible, and meaningful. It i
* [Federated Learning](/okf/en/glossary/federated-learning.md) - Federated Learning is a machine learning approach where models are trained across decentralized devices or servers holding local data samples, without exchangin
* [Fennec (Codename)](/okf/en/glossary/fennec-codename.md) - The internal codename for Anthropic's Claude Sonnet 5 model, following the tradition of using animal names for model development. The fennec fox is known for it
* [Fine-tuning](/okf/en/glossary/fine-tuning.md) - The process of taking a pre-trained machine learning model and further training it on a smaller, task-specific dataset to adapt its behavior for a particular us
* [Fleet Management (Code)](/okf/en/glossary/fleet-management-code.md) - A framework for applying code changes across a large number of repositories simultaneously, enabling efficient management of large-scale codebases. Used by comp
* [Full System Access](/okf/en/glossary/full-system-access.md) - The level of permissions granted to an AI assistant allowing it to interact with the operating system and hardware of a computer, including executing commands, 
* [Gemini 3 Flash](/okf/en/glossary/gemini-3-flash.md) - A specific version or model of Google's Gemini AI family, known for its speed (sub-500ms response times) and large context window (1M tokens).
* [Gemini 3 Pro](/okf/en/glossary/gemini-3-pro.md) - A large language model (LLM) developed by Google, emphasized for its coherence in multi-step reasoning chains.
* [Google Gemini](/okf/en/glossary/google-gemini.md) - A family of AI models developed by Google, designed for a wide range of tasks including text generation, coding, and multimodal processing.
* [GPT-4o](/okf/en/glossary/gpt-4o.md) - A large language model (LLM) created by OpenAI. GPT-4o is a multimodal conversational AI known for its conversational style and user perception of warmth.
* [GPT-5](/okf/en/glossary/gpt-5.md) - A hypothetical or future iteration of the GPT (Generative Pre-trained Transformer) series of large language models developed by OpenAI.
* [GPT-5.2](/okf/en/glossary/gpt-5-2.md) - A specific iteration of the Generative Pre-trained Transformer (GPT) model series developed by OpenAI, building upon GPT-5.1 with improvements in speed, accurac
* [GPT-5.2-Codex](/okf/en/glossary/gpt-5-2-codex.md) - A specialized variant of GPT-5.2 optimized for coding tasks, including code generation, debugging, security analysis, and working with diverse coding environmen
* [GPT-5.3-Codex](/okf/en/glossary/gpt-5-3-codex.md) - OpenAI's advanced coding model, an iteration of the GPT series specifically designed for code generation and completion of longer, end-to-end coding tasks, with
* [GraphRAG](/okf/en/glossary/graph-rag.md) - A sophisticated retrieval method using knowledge graphs to help AI understand complex relationships and context within large datasets better than standard searc
* [Grok 4](/okf/en/glossary/grok-4.md) - A large language model (LLM) developed by xAI, known for its advanced reasoning and performance.
* [GSD Framework](/okf/en/glossary/gsd-framework.md) - The GSD (Get Shit Done) Framework is a spec-driven development system for AI agents built on Claude Code. Uses 50 Markdown files, 6 slash commands, and 2 hooks 
* [Hook](/okf/en/glossary/hook.md) - A mechanism in Claude Code for triggering actions based on specific events or conditions within the system.
* [Hook System](/okf/en/glossary/hook-system-claude-code.md) - An event-driven mechanism in Claude Code plugin architecture that allows plugins to intercept and modify behavior at specific points in the agent workflow, such
* [Horizontal Connection (AI)](/okf/en/glossary/horizontal-connection-ai.md) - The connection between AI agents enabling communication, collaboration, and task delegation. ACP handles this dimension.
* [Hot-Reload Development](/okf/en/glossary/hot-reload-development.md) - A development workflow where code changes are applied instantly without restarting the application, enabling rapid iteration.
* [Human-AI Collaboration Design](/okf/en/glossary/human-ai-collaboration-design.md) - The process of designing systems and workflows that enable humans and AI to work together effectively, focusing on usability, trust, and shared goals.
* [Hybrid Attention](/okf/en/glossary/hybrid-attention.md) - An attention mechanism that combines different types of attention mechanisms (e.g., gated attention and delta net) to leverage the strengths of each and improve
* [Imagen 3](/okf/en/glossary/imagen-3-google-deepmind.md) - Google DeepMind third-generation text-to-image AI model that powers Google Whisk known for high photorealism and creative fidelity in image generation.
* [Imagen 3](/okf/en/glossary/imagen-3.md) - Google DeepMind's third-generation text-to-image AI model that powers Google Whisk, known for high photorealism and creative fidelity in image generation.
* [Inference Cost](/okf/en/glossary/inference-cost.md) - Inference cost refers to the financial expenditure incurred when operating an AI language model — the costs of processing every user request. Unlike training co
* [Inference Optimization](/okf/en/glossary/inference-optimization.md) - Inference optimization encompasses all techniques and strategies employed to improve the performance (latency, throughput) and/or cost efficiency of AI inferenc
* [Inference Scaling](/okf/en/glossary/inference-scaling.md) - Inference Scaling is the process of optimizing AI model deployment to handle a growing number of inference requests or increasing data volumes. This involves te
* [Inference-Time Compute](/okf/en/glossary/inference-time-compute.md) - Inference-Time Compute is a AI engineering concept in modern AI systems that improves the development and maintenance of AI-powered systems. It plays a key role
* [Injection Breakthroughs](/okf/en/glossary/injection-breakthroughs.md) - Instances where malicious or unintended external content injected into a prompt manages to bypass safety mechanisms and influence the LLM's behavior in an undes
* [Instruction/Data Separation](/okf/en/glossary/instruction-data-separation.md) - Separating trusted instructions from untrusted data.
* [Intelligent LLM Routing](/okf/en/glossary/intelligent-llm-routing.md) - Intelligent LLM Routing is a AI economics concept in modern AI systems that optimizes the cost-benefit equation of AI adoption and operation. It plays a key rol
* [Intent-Based Navigation](/okf/en/glossary/intent-based-navigation.md) - Intent-Based Navigation is a AI user experience concept in modern AI systems that shapes how users interact with and benefit from AI-powered features. It plays 
* [Interactive UI Components](/okf/en/glossary/interactive-ui-components.md) - Functional user interface elements (e.g., buttons, sliders, forms, dashboards) that allow users to directly interact with and manipulate data or trigger actions
* [iOS](/okf/en/glossary/ios.md) - Apple's mobile operating system, primarily used on iPhones and iPads.
* [JSON Mode](/okf/en/glossary/json-mode.md) - JSON Mode refers to the capability of a language model to provide its output in a structured JSON format. This is useful for applications requiring programmatic
* [JSON-RPC](/okf/en/glossary/json-rpc.md) - A remote procedure call protocol using JSON for data serialization, enabling systems to execute procedures over a network. Core communication method in MCP.
* [LLM-as-a-Judge Evaluations](/okf/en/glossary/llm-as-a-judge-evals.md) - LLM-as-a-Judge Evaluations is a AI infrastructure concept in modern AI systems that provides foundational capabilities for AI system deployment and operation. I
* [llms.txt](/okf/en/glossary/llms-txt.md) - A standard initiated by Jeremy Howard – a structured text file in a website's root directory that provides LLMs with optimized information about the website.
* [Local AI Inference](/okf/en/glossary/local-ai-inference.md) - Running AI model predictions directly on a user's device rather than sending data to cloud servers, providing privacy, lower latency, and no API costs.
* [Long-Context Model](/okf/en/glossary/long-context-model.md) - An AI language model capable of processing very large input sequences (hundreds of thousands to millions of tokens), enabling analysis of entire codebases or lo
* [Long-Term Memory Layers](/okf/en/glossary/long-term-memory-layers.md) - Long-Term Memory Layers is a AI infrastructure concept in modern AI systems that provides foundational capabilities for AI system deployment and operation. It p
* [macOS](/okf/en/glossary/macos.md) - Apple's operating system for Macintosh computers.
* [MCP (Model Context Protocol)](/okf/en/glossary/model-context-protocol.md) - A standardization effort, under the Linux Foundation, designed to provide a common framework for AI models and agents to exchange context, enabling interoperabi
* [MCP Servers](/okf/en/glossary/mcp-servers.md) - A server that enables Claude Code to integrate with external tools and services, allowing it to access and utilize their functionalities.
* [MCP Tasks (Async)](/okf/en/glossary/mcp-tasks-async.md) - MCP Tasks (Async) is a AI infrastructure concept in modern AI systems that provides foundational capabilities for AI system deployment and operation. It plays a
* [Mixture of Experts (MoE)](/okf/en/glossary/mixture-of-experts-moe.md) - Mixture of Experts (MoE) is a core AI technology concept in modern AI systems that represents fundamental technical capabilities powering modern AI applications
* [Mixture-of-Experts (MoE)](/okf/en/glossary/mixture-of-experts.md) - A neural network architecture that uses multiple 'expert' sub-networks. During inference, only a selected subset of these experts is activated, enabling a large
* [ML Engineer](/okf/en/glossary/ml-engineer.md) - A software engineer specializing in the development, deployment, and maintenance of machine learning models in production environments.
* [ML Engineers](/okf/en/glossary/ml-engineers.md) - Machine Learning Engineers who design, build, and deploy ML models and systems. They bridge the gap between data science and software engineering.
* [Model Distillation](/okf/en/glossary/model-distillation.md) - A technique where a smaller, faster AI model is trained to replicate the capabilities of a larger model, enabling cost-effective deployment while maintaining hi
* [Model Quantization](/okf/en/glossary/model-quantization.md) - Model Quantization is a technique to reduce the memory footprint and computational requirements of AI models by representing weights and activations with lower 
* [Model-agnostic](/okf/en/glossary/model-agnostic.md) - Refers to a system or software that is designed to work with various AI language models, rather than being specifically tied to one particular model.
* [Model-Agnostic](/okf/en/glossary/model-agnostic-ai.md) - A system design approach where the AI framework works with any language model provider rather than being locked to a specific one. Allows switching between GPT-
* [Modular Extension System](/okf/en/glossary/modular-extension-system.md) - A system that allows users to customize and extend the functionality of a software application (like Claude Code) by adding, removing, or modifying self-contain
* [Multi-Agent Coding](/okf/en/glossary/multi-agent-coding.md) - The process of developing software using multiple AI agents that work in parallel or sequentially to complete coding tasks.
* [Multi-Agent Coding Workflow](/okf/en/glossary/multi-agent-coding-workflow.md) - A software development workflow where multiple AI agents work in parallel on different coding tasks, coordinated through a central interface like Codex App.
* [Multi-Agent Orchestration](/okf/en/glossary/multi-agent-orchestration.md) - The coordination of multiple specialized AI agents working together as a digital team to solve complex, cross-departmental problems.
* [Multi-Agent Platform](/okf/en/glossary/multi-agent-platform-ai.md) - A software environment orchestrating multiple AI agents with different capabilities to collaborate on complex tasks. GitHub Agent HQ exemplifies this by assigni
* [Multi-Agent Platform](/okf/en/glossary/multi-agent-platform.md) - A platform that allows developers to use and manage multiple AI agents, often from different providers, within a unified environment.
* [Multi-Agent PR Review](/okf/en/glossary/multi-agent-pr-review.md) - A code review approach that dispatches multiple AI agents in parallel to analyze a pull request from different perspectives simultaneously. Unlike single-pass t
* [Multi-Agent Workflow](/okf/en/glossary/multi-agent-workflow.md) - A system where multiple AI agents collaborate and coordinate to achieve a complex goal, often involving handoffs and dependencies between agents.
* [Multi-Modal Feedback Loops](/okf/en/glossary/multi-modal-feedback-loops.md) - Multi-Modal Feedback Loops is a AI user experience concept in modern AI systems that shapes how users interact with and benefit from AI-powered features. It pla
* [Multimodal Model](/okf/en/glossary/multimodal-ai-model.md) - An AI model that processes and generates multiple data types — text, images, audio, video — within a single architecture. Models like GPT-4o and Gemini understa
* [Multimodal Model](/okf/en/glossary/multimodal-model.md) - An AI model capable of processing and integrating information from multiple modalities, such as text, images, and audio.
* [n8n](/okf/en/glossary/n8n.md) - An open-source, node-based workflow automation platform that allows users to connect various applications and services to automate tasks.
* [n8n-MCP server](/okf/en/glossary/n8n-mcp-server.md) - An open-source server acting as a bridge between n8n and AI models (like Claude Code) via the Model Context Protocol (MCP), enabling AI to control and automate 
* [Nano Banana Pro](/okf/en/glossary/nano-banana-pro.md) - An ultra-efficient open-source image generator from Google released in December 2025 that produces high-quality images with minimal resource consumption.
* [Natural Language Autoencoder (NLA)](/okf/en/glossary/natural-language-autoencoders.md) - A natural language autoencoder (NLA) is an interpretability technique from AI safety research that translates a language model's internal activations into a pla
* [Natural Language Workflow Creation](/okf/en/glossary/natural-language-workflow-creation.md) - The process of defining and creating automated workflows using natural language prompts, which are then translated into executable instructions by an AI system.
* [Natural-Language-Driven](/okf/en/glossary/natural-language-driven.md) - A development approach where natural language is used as the primary input for software creation, interpreted and executed by AI.
* [NemoClaw](/okf/en/glossary/nemoclaw.md) - NemoClaw is Context Studios' internal agent framework, developed specifically for creating and managing AI agent pipelines in the content and marketing domain. 
* [NPU Optimization](/okf/en/glossary/npu-optimization.md) - NPU or Neural Processing Unit Optimization refers to techniques for maximizing the performance of dedicated AI accelerator chips. NPUs are specialized hardware 
* [Observe-Think-Act loop](/okf/en/glossary/observe-think-act-loop.md) - The fundamental cycle of an AI agent, where it perceives its environment (Observe), processes the information and decides on a course of action (Think), and the
* [Ollama Ecosystem](/okf/en/glossary/ollama-ecosystem.md) - The Ollama Ecosystem refers to the tools, models, and community around Ollama, an open-source platform for running large language models locally. It simplifies 
* [Open-weight Model](/okf/en/glossary/open-weight-model.md) - A language model whose weights (the learned parameters) are publicly available for download and use.
* [Open-Weight Model](/okf/en/glossary/open-weight.md) - An open-weight model is a type of artificial intelligence model where the trained parameters (weights) are publicly released for download, inspection, fine-tuni
* [OpenAI Apps SDK](/okf/en/glossary/openai-apps-sdk.md) - A UI framework that facilitates the creation of cross-platform AI applications by providing tools to translate between different AI platforms and MCP servers.
* [OpenAI Codex](/okf/en/glossary/openai-codex-agent.md) - OpenAI's cloud-based AI coding agent that executes code, manages repos, and handles dev tasks autonomously in a sandboxed environment with GitHub integration.
* [OpenAI Codex](/okf/en/glossary/openai-codex.md) - A model from OpenAI that translates natural language into code; the foundation for Copilot.
* [OpenAI Connectors](/okf/en/glossary/openai-connectors.md) - Wrappers developed by OpenAI that facilitate integration between AI agents and commonly used applications and services like Google Drive, Slack, and Notion, oft
* [OpenClaw](/okf/en/glossary/openclaw.md) - An open-source framework for creating and running autonomous AI agents that can operate across multiple messaging platforms.
* [Operational Foundation](/okf/en/glossary/operational-foundation.md) - The core systems and processes upon which an organization runs its daily activities, now increasingly reliant on AI agents.
* [Output Contract](/okf/en/glossary/output-contract.md) - A clearly defined structure and format for the output generated by an AI agent, ensuring consistency and predictability for downstream processing or human consu
* [Parameters (Model Parameters)](/okf/en/glossary/model-parameters.md) - The adjustable weights within a neural network that are learned during training. They determine the model's ability to map inputs to outputs.
* [Persistent Memory (AI)](/okf/en/glossary/persistent-memory-ai.md) - The ability of an AI system to retain information across sessions and conversations, building long-term context about users and tasks. Enables continuity and pe
* [Personal Intelligence](/okf/en/glossary/personal-intelligence-ai.md) - AI systems deeply integrated into individual users' lives, learning preferences and proactively assisting with daily tasks and decisions. A vision articulated b
* [Personal Intelligence](/okf/en/glossary/personal-intelligence.md) - Google's initiative to expand its Gemini model with capabilities tailored to individual user needs and preferences, providing a highly personalized AI experienc
* [Platform Consolidation](/okf/en/glossary/platform-consolidation-ai.md) - The trend of replacing multiple specialized software tools with fewer comprehensive platforms. In the AI era, AI super apps absorb functions of separate SaaS pr
* [Platform Consolidation](/okf/en/glossary/platform-consolidation.md) - The trend of merging multiple functionalities and services into a single, unified platform, often powered by AI, reducing the need for disparate applications.
* [Privacy-Preserving Inference](/okf/en/glossary/privacy-preserving-inference.md) - Privacy-Preserving Inference is a regulatory compliance concept in modern AI systems that addresses legal and regulatory requirements for AI deployment. It play
* [Private Cloud Compute](/okf/en/glossary/apple-private-cloud-compute.md) - Apple cloud infrastructure processing AI requests with hardware-level privacy guarantees. Data in secure enclaves, never stored.
* [Private Cloud Compute](/okf/en/glossary/private-cloud-compute.md) - Apple's technology that processes sensitive data locally on device or through private, secure servers to preserve user privacy while utilizing AI capabilities. 
* [Proactive Communication](/okf/en/glossary/proactive-communication.md) - The ability of an AI assistant to initiate communication with a user based on triggers, events, or learned preferences, rather than waiting for explicit request
* [Programmatic Access](/okf/en/glossary/programmatic-access.md) - Accessing a software system or service through code (e.g., using an API) rather than a graphical user interface.
* [Prompt Caching](/okf/en/glossary/prompt-caching.md) - A technique that stores frequently used context in an AI model's memory, drastically reducing latency and costs for repetitive queries.
* [Prompt Injection Defense](/okf/en/glossary/prompt-injection-defense.md) - Prompt Injection Defense is a AI safety concept in modern AI systems that ensures AI systems operate within safe boundaries and produce reliable outputs. It pla
* [Prompt Re-injection](/okf/en/glossary/prompt-re-injection.md) - The process of feeding the same initial prompt back into an AI model to encourage it to continue working on a task, building upon previous iterations.
* [Prompt Template](/okf/en/glossary/prompt-template.md) - Reusable prompt structures with placeholders for dynamic content. Enable consistent outputs across different inputs. Best practices: Clear role definition, stru
* [Provenance](/okf/en/glossary/provenance.md) - The documentation of the origin and history of a piece of data, including where it came from, how it was derived, and who has modified it.  In the context of LL
* [Quantization (AI)](/okf/en/glossary/quantization-ai.md) - A technique that reduces the precision of an AI model's numerical weights (e.g., from 32-bit to 4-bit), dramatically shrinking model size and memory requirement
* [Ralph Wiggum Plugin](/okf/en/glossary/ralph-wiggum-plugin-claude-code.md) - A Claude Code plugin that enables fully autonomous AI development by automatically accepting all tool calls and permissions allowing Claude Code to work without
* [Ralph Wiggum Plugin](/okf/en/glossary/ralph-wiggum-plugin.md) - A Claude Code plugin that enables fully autonomous AI development by automatically accepting all tool calls and permissions, allowing Claude Code to work withou
* [Ralph Wiggum Technique](/okf/en/glossary/ralph-wiggum-technique.md) - An autonomous development methodology for AI coding assistants, invented by Geoffrey Huntley. Uses a Stop Hook that intercepts Claude's exit attempts and repeat
* [React Server Components](/okf/en/glossary/react-server-components.md) - A React feature that allows components to run on the server instead of the client, improving performance and reducing client-side JavaScript.
* [Refactoring](/okf/en/glossary/refactoring.md) - The process of restructuring existing computer code—changing the factoring—without changing its external behavior.
* [Resources (MCP)](/okf/en/glossary/mcp-resources.md) - Structured data that an AI assistant can access through the Model Context Protocol (MCP), like database schemas or documentation.
* [Responses API](/okf/en/glossary/responses-api.md) - OpenAI's API for generating structured responses from AI models, supporting tool use, function calling, and multi-step reasoning workflows.
* [Revenue Validation](/okf/en/glossary/revenue-validation.md) - Confirming a product idea can generate paying customers before committing significant development resources.
* [Runtime](/okf/en/glossary/runtime.md) - The environment in which a computer program or AI agent is executed, encompassing the software and hardware resources needed for its operation.
* [SaaS](/okf/en/glossary/saas.md) - Software as a Service. Software provided as a service over the Internet.
* [Sandboxed Container](/okf/en/glossary/sandboxed-container.md) - A secure, isolated environment for running applications that limits access to system resources and prevents interference with other processes. Critical for ente
* [Sandboxed Web Access](/okf/en/glossary/sandboxed-web-access.md) - A Claude Code security configuration that restricts the AI agent's browsing capabilities to a defined whitelist of domains. Instead of unrestricted internet acc
* [Scaling AI](/okf/en/glossary/scaling-ai.md) - The process of increasing the size, complexity, and resources allocated to AI models and systems, often involving expanding the training dataset, model paramete
* [Schema Validation](/okf/en/glossary/schema-validation.md) - The process of verifying that the output of a language model conforms to a predefined structure or format (schema).
* [Seat](/okf/en/glossary/seat.md) - In the context of software licensing, a 'seat' represents a single user or user account authorized to access and use the software. SaaS pricing is often based o
* [Seedance 2.0](/okf/en/glossary/seedance-2-0.md) - Seedance 2.0 is a multimodal AI video generation model developed by ByteDance, the Beijing-based technology company best known for TikTok. Released in 2025, See
* [Self-hosted](/okf/en/glossary/self-hosted.md) - A software or service that is hosted on the user's own infrastructure, giving them greater control over data and resources compared to cloud-based solutions.
* [Semantic Caching](/okf/en/glossary/semantic-caching.md) - A technique that stores AI responses for similar (not just identical) queries, allowing the system to serve answers instantly without incurring new API costs.
* [SEP (MCP Enhancement Proposal)](/okf/en/glossary/mcp-enhancement-proposal-sep.md) - A design document that provides information to the MCP community or describes a new feature for the Model Context Protocol. SEP governance means breaking change
* [Session Continuity](/okf/en/glossary/session-continuity.md) - Session continuity refers to the ability of an AI agent or system to maintain state, context, and progress across interruptions, restarts, or session changes. S
* [Skill](/okf/en/glossary/skill.md) - A modular, reusable component designed to perform a specific task within Claude Code. Skills can be triggered by user input or other events.
* [Skill Definition (in YAML format)](/okf/en/glossary/skill-definition-yaml.md) - Defining the specific capabilities and functions of an AI agent using YAML, a human-readable data-serialization language. This outlines what the agent can do an
* [Skill Hot-Reload](/okf/en/glossary/skill-hot-reload.md) - The ability to automatically update and reload skills in a development environment without requiring a complete restart of the system.
* [Skills](/okf/en/glossary/skills.md) - Auto-invoked capabilities of Claude Code triggered by context, allowing it to perform tasks without explicit user commands.
* [Skills API](/okf/en/glossary/skills-api.md) - An Application Programming Interface (API) that allows developers to create, manage, and integrate reusable sets of instructions ('Skills') into AI platforms, e
* [Skills System](/okf/en/glossary/skills-system.md) - A modular framework within an AI assistant that allows users to extend its functionality through installable plugins or modules.
* [Skills System (AI)](/okf/en/glossary/skills-system-ai.md) - A modular architecture where AI capabilities are organized as discrete, pluggable skill modules that can be added, removed, or updated independently. Used in sy
* [Software 3.0](/okf/en/glossary/software-30.md) - Software 3.0 refers to the paradigm where AI models become the primary way software behaves, moving beyond traditional code (1.0) and neural networks trained on
* [Software for One](/okf/en/glossary/software-for-one.md) - Personalized software tools quickly built for individual use, often leveraging AI for rapid prototyping and automation.
* [Sovereign Cloud AI (GAIA-X)](/okf/en/glossary/sovereign-cloud-ai-gaia-x.md) - Sovereign Cloud AI (GAIA-X) is a regulatory compliance concept in modern AI systems that addresses legal and regulatory requirements for AI deployment. It plays
* [Spatial Intelligence](/okf/en/glossary/spatial-intelligence.md) - The capability of AI to perceive, reason about, and interact with 3D spaces, bridging the gap between digital intelligence and physical reality.
* [Startup Ecosystem](/okf/en/glossary/startup-ecosystem-ai.md) - The interconnected network of startups, investors, accelerators, universities, government programs, and support organizations that foster entrepreneurship and i
* [Stop Hook](/okf/en/glossary/stop-hook.md) - A mechanism that intercepts the normal termination or exit behavior of an AI model, allowing for modifications or continued operation before the model concludes
* [Streamable HTTP Transport](/okf/en/glossary/streamable-http-transport.md) - Streamable HTTP Transport is a AI infrastructure concept in modern AI systems that provides foundational capabilities for AI system deployment and operation. It
* [Subagents](/okf/en/glossary/subagents.md) - Smaller, specialized AI agents that work together within a larger AI agent system to accomplish complex tasks.
* [Swift Assist](/okf/en/glossary/swift-assist.md) - Apple's AI-powered coding assistant integrated into Xcode that helps developers write, understand, and debug Swift code using large language models.
* [Synthetic Data Generation](/okf/en/glossary/synthetic-data-generation.md) - Synthetic Data Generation involves creating artificial data that mimics the statistical properties of real-world data. This is often used to augment or replace 
* [Tech Stack](/okf/en/glossary/tech-stack-ai.md) - The complete collection of technologies used to build and run a software application, including programming languages, frameworks, libraries, databases, and clo
* [Technical Debt Tsunami](/okf/en/glossary/technical-debt-tsunami.md) - A metaphor describing the overwhelming accumulation of technical debt resulting from rushed or poorly planned development practices, particularly when using AI-
* [Terminal Workflow](/okf/en/glossary/terminal-workflow.md) - The set of tasks, commands, and processes a developer or user executes within a command-line interface (terminal) for software development, system administratio
* [Test-Time Compute](/okf/en/glossary/test-time-compute.md) - Test-Time Compute refers to the computational resources required to run inference or make predictions using a trained AI model. Efficient test-time compute is c
* [Test-Time Scaling](/okf/en/glossary/test-time-scaling.md) - The practice of dedicating more computational power at the moment of generating an answer (inference) rather than just during training, allowing the model to 't
* [Time-to-First-Token (TTFT)](/okf/en/glossary/time-to-first-token.md) - The latency measured from when a user sends a prompt to a language model until the first token of the response begins streaming back. TTFT is the most important
* [Token Budget](/okf/en/glossary/token-budget.md) - The limited number of tokens (text units) that can be included in a language model's input context due to cost, performance, or model limitations. This budget c
* [Token Input Context](/okf/en/glossary/token-input-context.md) - The maximum number of tokens (units of text) that an AI model can process as input in a single request.
* [Token Yield Optimization](/okf/en/glossary/token-yield-optimization.md) - Token Yield Optimization is a AI economics concept in modern AI systems that optimizes the cost-benefit equation of AI adoption and operation. It plays a key ro
* [Tokens (in LLMs)](/okf/en/glossary/tokens-in-llms.md) - The basic units of text that LLMs process, typically words or parts of words. Token consumption refers to the number of tokens used for both input and output, i
* [Tool Use / Function Calling](/okf/en/glossary/tool-use-function-calling.md) - Tool Use / Function Calling is a AI infrastructure concept in modern AI systems that provides foundational capabilities for AI system deployment and operation. 
* [Tool Use in AI](/okf/en/glossary/tool-use-in-artificial-intelligence.md) - The capability of AI models to interact with external software tools APIs and services during inference to gather information or extend capabilities.
* [Tools (MCP)](/okf/en/glossary/mcp-tools.md) - Executable actions that an AI assistant can trigger through the Model Context Protocol (MCP), such as writing a file or calling an API.
* [Turbopack](/okf/en/glossary/turbopack.md) - A high-performance build tool for JavaScript and TypeScript, designed as a successor to Webpack. Notably faster build times through caching.
* [Typicality Bias](/okf/en/glossary/typicality-bias.md) - The systematic human preference for 'typical' texts over unusual ones – a well-documented phenomenon in cognitive psychology. Measured at α = 0.57±0.07 in LLM a
* [Unified Playground](/okf/en/glossary/unified-playground.md) - A consolidated interface or environment that provides access to multiple AI models and tools, enabling users to experiment, compare, and utilize different model
* [val_bpb (Validation Bits Per Byte)](/okf/en/glossary/val-bpb.md) - A performance metric for language models measuring how efficiently a model compresses validation data. Calculated as bits per byte of text, lower values indicat
* [Veo 3.1](/okf/en/glossary/veo-3-1-google-video-model.md) - Google latest video generation AI model that powers Google Flow offering high-quality video synthesis with support for complex transitions and native audio gene
* [Veo 3.1](/okf/en/glossary/veo-3-1.md) - Google's latest video generation AI model that powers Google Flow, offering high-quality video synthesis with support for complex transitions and native audio g
* [Verbalized Sampling](/okf/en/glossary/verbalized-sampling-technique.md) - A technique to combat mode collapse by explicitly instructing the AI model through natural language prompts to generate diverse outputs rather than relying on t
* [Vertical Connection](/okf/en/glossary/vertical-connection.md) - In the context of AI agents, a vertical connection refers to the link between an agent and the external tools, databases, and APIs it uses to perform tasks.
* [Vertical Connection (AI)](/okf/en/glossary/vertical-connection-ai.md) - The connection between an AI agent and the tools, databases, and APIs it needs to access external information and perform tasks. In the AI protocol landscape, M
* [Vibe Coding](/okf/en/glossary/vibe-coding-definition.md) - A development approach where AI generates code from high-level human guidance.
* [Vibe Coding Hangover](/okf/en/glossary/vibe-coding-hangover.md) - The negative consequences after extensive vibe coding: accumulated technical debt, unmaintainable code, security vulnerabilities, and the realization that AI-ge
* [Vulnerability Scanning](/okf/en/glossary/vulnerability-scanning.md) - The automated process of identifying security weaknesses and potential vulnerabilities in software, networks, or systems. Modern approaches range from rule-base
* [Wafer-Scale Engine (WSE)](/okf/en/glossary/wafer-scale-engine.md) - A revolutionary chip architecture developed by Cerebras Systems where an entire 300mm silicon wafer is used as a single processor, rather than being cut into hu
* [Workflow Redesign](/okf/en/glossary/workflow-redesign.md) - Re-engineering existing business processes to incorporate and optimize the use of AI agents, often resulting in significant efficiency gains.
* [World Models](/okf/en/glossary/world-models.md) - AI systems that develop a grounded understanding of physical and causal laws, allowing them to predict outcomes in virtual and real environments.
* [Xcode](/okf/en/glossary/xcode.md) - Apple's Integrated Development Environment (IDE) for developing software for macOS, iOS, watchOS, and tvOS.
* [Xcode Previews](/okf/en/glossary/xcode-previews.md) - A feature within Xcode that allows developers to see a real-time visual representation of their UI as they code.
* [Agent Orchestration](/okf/en/glossary/agent-orchestration.md) - Agent orchestration refers to the coordination of multiple AI agents by a central orchestrator agent or orchestration system to solve complex tasks that individ
* [Agent Reliability](/okf/en/glossary/agent-reliability.md) - Agent reliability refers to the degree to which an AI agent consistently and correctly completes desired tasks without unexpected failures, runaway behavior, or
* [Agentic Coding](/okf/en/glossary/agentic-coding-ai.md) - Agentic coding is an emerging paradigm in software development where AI agents autonomously write, test, debug, and refactor code with minimal human interventio
* [Agentic Coding Tools](/okf/en/glossary/agentic-coding-tools.md) - Software platforms like Claude Code that enable AI agents to autonomously write, test, and review code. Unlike simple autocomplete, agentic coding tools execute
* [AI Coding Agent](/okf/en/glossary/ai-coding-agent.md) - An AI-powered system designed to autonomously generate, modify, and deploy code, integrating with development workflows like CI/CD pipelines and version control
* [AI Coding Desktop App](/okf/en/glossary/ai-coding-desktop-app.md) - A standalone desktop application designed for AI-assisted software development, offering agent management, task monitoring, and integrated development workflows
* [AI Computer Use](/okf/en/glossary/computer-use-ai.md) - AI computer use refers to the ability of AI agents to directly operate a computer — moving the mouse, clicking, typing text, reading screen content, and accessi
* [AI Inference](/okf/en/glossary/ai-inference.md) - AI inference is the process by which a trained machine learning model processes new input data to generate predictions, text, images, or other outputs. Unlike t
* [AI Model Ping-Pong](/okf/en/glossary/ai-model-ping-pong.md) - AI Model Ping-Pong is a AI economics concept in modern AI systems that optimizes the cost-benefit equation of AI adoption and operation. It plays a key role in 
* [AI Stock Selloff](/okf/en/glossary/ai-stock-selloff.md) - An AI stock selloff refers to a significant decline in the share prices of AI-related companies. In February 2026, the S&P 500 Software & Services Index experie
* [Async Agentic Coding](/okf/en/glossary/async-agentic-coding.md) - A development workflow where an AI coding agent runs autonomously on your local machine, executing tasks over an extended period without requiring constant deve
* [Batch Inference](/okf/en/glossary/batch-inference.md) - Batch inference is the process of collecting multiple AI requests and processing them together as a group, rather than handling each individually and immediatel
* [Benchmark Contamination](/okf/en/glossary/benchmark-contamination.md) - Benchmark contamination refers to the problem where evaluation data — the questions and answers comprising a benchmark — appears in a model's training data, eit
* [Coding Agent](/okf/en/glossary/coding-agent.md) - An AI system that goes beyond code completion to autonomously perform complex software engineering tasks like implementing features, fixing bugs, running tests,
* [Context Window](/okf/en/glossary/context-window-llm.md) - The context window is the maximum amount of text — measured in tokens — that a large language model can process and attend to in a single inference call. Tokens
* [EHR](/okf/en/glossary/ehr-electronic-health-record.md) - Electronic Health Record — digital patient medical chart. AI agents increasingly integrate with EHR systems for automated documentation and clinical decision su
* [Evidence Packs](/okf/en/glossary/evidence-packs.md) - Curated document bundles provided to the AI model as a verified factual foundation for complex tasks, significantly reducing hallucinations.
* [GGUF Format](/okf/en/glossary/gguf-format.md) - GGUF is a file format for storing quantized large language models, designed for efficient loading and inference. It replaced the older GGML format and is widely
* [GLM-5](/okf/en/glossary/glm-5.md) - GLM-5 is a large language model developed by Zhipu AI, a Beijing-based AI research company, featuring approximately 744 billion parameters — making it one of th
* [Human-in-the-Loop (HITL)](/okf/en/glossary/human-in-the-loop.md) - Human oversight integrated into AI decision-making.
* [Inference Chip](/okf/en/glossary/inference-chip.md) - An inference chip is a specialized semiconductor processor optimized for efficiently running AI models during inference. Unlike general-purpose CPUs or training
* [Injection Attack (LLM)](/okf/en/glossary/injection-attack-llm.md) - Malicious instructions in input to manipulate LLM behavior.
* [JSON-RPC](/okf/en/glossary/json-rpc-protocol.md) - A lightweight remote procedure call protocol encoded in JSON used as the communication layer in the Model Context Protocol for standardized message exchange bet
* [Medical Coding](/okf/en/glossary/medical-coding-ai.md) - Translating medical diagnoses into standardized codes (ICD-10/11, CPT). AI agents automate this error-prone process, reducing claim rejections.
* [Mixture-of-Experts (MoE)](/okf/en/glossary/moe-architecture.md) - Mixture-of-Experts (MoE) is a neural network architecture in which a model consists of multiple specialized sub-networks called experts, paired with a learned g
* [Model Retirement](/okf/en/glossary/model-retirement.md) - Model retirement is the process by which AI companies deprecate and discontinue older AI models, redirecting users to newer versions. OpenAI's retirement of GPT
* [Moonshot AI](/okf/en/glossary/moonshot-ai.md) - A Chinese AI company that developed the Kimi series of language models, known for pioneering ultra-long context windows and competitive open-source models that 
* [Multi-Agent Communication](/okf/en/glossary/multi-agent-communication.md) - Multi-agent communication encompasses the protocols, mechanisms, and patterns through which multiple AI agents interact, exchange information, and coordinate ta
* [Multimodal AI](/okf/en/glossary/multimodal-ai.md) - Multimodal AI refers to artificial intelligence systems capable of processing, understanding, and generating information across multiple data modalities — inclu
* [NVIDIA Blackwell](/okf/en/glossary/nvidia-blackwell.md) - NVIDIA Blackwell is NVIDIA's latest-generation AI GPU architecture, named after mathematician David Harold Blackwell. Unveiled at GTC 2024 with further announce
* [NVIDIA Vera Rubin](/okf/en/glossary/nvidia-vera-rubin.md) - NVIDIA Vera Rubin is the next-generation GPU architecture following Blackwell, announced by Jensen Huang at GTC 2026 and planned for 2026/2027 deployment. Named
* [Phase Budget](/okf/en/glossary/phase-budget.md) - A phase budget is an explicitly defined time limit or token limit for a single phase within an AI agent workflow. The concept originates from the GSD Framework 
* [Real-Time Inference](/okf/en/glossary/real-time-inference.md) - Real-time inference is the immediate processing of AI requests with minimal latency, typically in the range of milliseconds to a few seconds. Unlike batch infer
* [Research Preview](/okf/en/glossary/research-preview.md) - A pre-release software version available to limited users for testing before official launch. Common in AI product releases.
* [SaaS Sprawl](/okf/en/glossary/saas-sprawl.md) - SaaS sprawl refers to the uncontrolled growth of software-as-a-service subscriptions within an organization. The average company uses 130+ SaaS tools, with 25-3
* [Sandboxed Iframe](/okf/en/glossary/sandboxed-iframe.md) - A sandboxed iframe is a restricted HTML container that isolates embedded content from the parent page for security. In the context of MCP Apps, sandboxed iframe
* [Semantic Router](/okf/en/glossary/semantic-router.md) - A lightweight layer that classifies user intent and routes it to the most efficient sub-agent or model, saving time and money.
* [Small Language Model (SLM)](/okf/en/glossary/small-language-model.md) - An AI language model with relatively few parameters (typically under 10B) designed for efficient local deployment on consumer hardware while maintaining useful 
* [Spec-Driven Development](/okf/en/glossary/spec-driven-development.md) - Spec-Driven Development is a AI engineering concept in modern AI systems that improves the development and maintenance of AI-powered systems. It plays a key rol
* [Spec-Driven Scaffolding](/okf/en/glossary/spec-driven-scaffolding.md) - Spec-driven scaffolding is the practice of controlling AI agents not through free-form prompts but through structured, machine-readable specifications — similar
* [Speculative Decoding](/okf/en/glossary/speculative-decoding.md) - An optimization technique where a small, fast model predicts the next few tokens, and a larger model only verifies them, drastically increasing speed.
* [Text-to-Video](/okf/en/glossary/text-to-video.md) - Text-to-video is a category of generative AI technology in which models produce video sequences directly from natural language descriptions, without traditional
* [Tokens Per Second (TPS)](/okf/en/glossary/tokens-per-second.md) - Tokens Per Second (TPS) is the primary throughput metric for evaluating AI language model inference performance. It measures how many tokens a model generates p
* [Vendor Lock-In (AI)](/okf/en/glossary/vendor-lock-in-ai.md) - Vendor lock-in in AI refers to the dependency on a single AI provider's models, tools, and ecosystem, making it costly to switch. GitHub's Agent HQ addresses th
* [YAML Frontmatter](/okf/en/glossary/yaml-frontmatter.md) - A metadata block at the top of a file written in YAML format, commonly used to configure AI agent skills, blog posts, and documentation.