Glossary
A comprehensive glossary of open source and developer terminology
A
Andrej Karpathy
Co-founder of OpenAI and former AI leader at Tesla. Coined 'vibe coding' in a February 2, 2025 post on X.
Agentic Coding
The evolution of vibe coding where AI agents work autonomously on complex, multi-step development tasks with minimal human intervention.
AI Coding Benchmarks
Standardized tests for evaluating AI coding capability, including SWE-bench, HumanEval, and MBPP.
Andrew Ng
Influential AI researcher (Stanford, Coursera, DeepLearning.AI) who has criticized the term 'vibe coding' as misleading about the engineering involved.
Agentic Engineering
A term proposed by Andrej Karpathy as a more professional alternative to 'vibe coding,' emphasizing engineering rigor in guiding AI agents.
Agent Hooks
In Kiro, autonomous background tasks triggered by events like file save. Agents execute predefined prompts without manual intervention.
AI-Assisted Development
The broad category encompassing all forms of AI involvement in software creation, from autocomplete to vibe coding to fully autonomous agentic coding.
Agent Mode
A feature in Cursor and GitHub Copilot that gives the AI greater autonomy, enabling file-wide edits, test generation, and CLI commands through natural language.
AI Pair Programming
The practice of working alongside an AI assistant as if it were a human pair programming partner, where the AI writes code and the human reviews.
AAIF (Agentic AI Foundation)
A directed fund under the Linux Foundation created in December 2025, co-founded by Anthropic, Block, and OpenAI, that governs the MCP standard.
AI Code Generation
The process of using AI models to produce source code from natural language descriptions, existing code patterns, or specifications.
Antigravity (Google)
Google's AI editor, set to benefit from the MCP standard adopted in Xcode and other tools.
B
Bolt.new
Rapid full-stack app generation from StackBlitz that spins up development environments instantly and generates production-ready apps with minimal prompting.
Base44
AI-powered platform for creating scalable applications with secure workflows and built-in collaboration, focusing on code structure and clean execution.
Background Agents
AI agents that work asynchronously while the developer does other things, allowing tasks to be started, left, and reviewed later.
C
Claude Code
Anthropic's command-line agentic coding tool that understands entire repositories, makes multi-file changes, and runs tests. Defined the agentic coding category.
Claude Desktop
Anthropic's desktop application for interacting with Claude. Acts as an MCP host, allowing connection to local MCP servers for file access and database queries.
Context7
MCP server that fetches up-to-date library and framework documentation so AI tools don't suggest deprecated APIs.
Cursor Rules / .cursorrules
Configuration files that provide persistent context to Cursor about your project's tech stack, coding conventions, and preferences.
Cursor
The most popular AI-powered code editor. A VS Code fork with deep AI integration including chat-based editing, Agent mode, and MCP support.
CLAUDE.md
A project-level configuration file for Claude Code that provides context about your project, coding standards, preferred workflows, and tools.
Claude Opus 4.6
Anthropic's most advanced AI model (February 2026). Part of the Claude 4.5 family with enhanced coding skills including planning, code review, and debugging.
Claude Sonnet 4.5
Anthropic's midsize model that balances capability and speed. Widely used in Cursor and other AI coding tools.
Claude Agent SDK
The underlying harness that powers Claude Code. Apple's Xcode 26.3 natively integrates with the Claude Agent SDK.
Claude Cowork
Anthropic's desktop tool for non-developers to delegate complex tasks to AI agents, including research, document creation, and data analysis.
Cline
An open-source AI coding agent that runs in VS Code, supporting multiple AI providers with an emphasis on transparency and user control.
Context Window
The maximum amount of text an AI model can process in a single interaction. Larger context windows allow models to understand bigger codebases.
Context Engineering
The skill of providing AI tools with the right information to produce accurate, relevant code. Considered the most important skill in AI-assisted development.
F
Figma MCP
Figma's official Dev Mode MCP server exposing design layer structure, auto-layout, variants, and design tokens to AI coding tools.
Fleet
A lightweight editor from JetBrains with built-in AI support. Fast, simple, and ideal for multi-language work.
Function Calling / Tool Use
A feature of AI models that allows them to invoke predefined functions based on user requests. MCP standardizes how function calling works across models and tools.
Firebase Studio
Google's full-stack development environment for building apps with Gemini. Supports iterative refinement and deployment.
G
GPT-5.2-Codex
OpenAI's most powerful coding model as of early 2026. Powers the Codex macOS app and claims to be the strongest model for sophisticated agentic coding work.
GitHub MCP
Official MCP server for searching code, browsing PRs, managing issues, and accessing repository data through AI tools.
GitHub Copilot
The most widely adopted AI coding assistant with 1M+ developers and 3B+ lines generated. Has evolved from autocomplete to chat to agent mode.
GitLab MCP
MCP server for CI/CD pipeline status, merge request context, and issue tracking within GitLab environments.
Git Worktrees
A Git feature used in parallel agent execution to let multiple agents work on the same codebase simultaneously on separate branches.
Google AI Studio
Google's platform for single-prompt app generation with one-click deployment to Cloud Run. The fastest path from concept to live app in Google's ecosystem.
Gemini Code Assist
Google's AI coding assistant powered by Gemini models. Integrates into existing IDEs for code suggestions, debugging, and generation.
L
Lovable
Chat-driven full-stack app builder with one-click deployment. Generates UI, backend logic, data models, and hosting from natural language.
Language Server Protocol (LSP)
An existing protocol that standardizes how programming language features work across different editors. MCP was partially inspired by LSP's approach.
M
MCP (Model Context Protocol)
An open standard created by Anthropic that connects AI tools to external data sources and services through a universal interface. Built on JSON-RPC 2.0.
MCP Server
A lightweight program that exposes specific capabilities to AI applications through the MCP standard. Over 16,000 MCP servers exist as of early 2026.
MCP Client
A component built into an AI host application that manages connections to MCP servers. Developers don't interact with clients directly.
MCP Host
The user-facing application that contains the MCP client, such as Claude Desktop, Cursor, Xcode, or any MCP-compatible IDE.
Magic UI MCP
MCP server exposing the Magic UI React/Tailwind component library for generating production-ready animated components and marketing sections.
MCP Apps
Interactive UI elements rendered directly inside AI host applications, beyond text-only responses. An evolution of MCP-UI being incorporated into the MCP standard.
Manus
End-to-end project automation platform that handles research, planning, coding, and deployment. The most autonomous of the full-stack platforms.
Multi-Agent Systems
Architectures where multiple AI agents collaborate on different aspects of the same project, such as frontend, backend, and testing.
O
Open Source Sustainability (AI Impact)
The growing concern that vibe coding threatens open source projects, as AI agents consume open source libraries at scale without contributing back.
OpenAI Codex
OpenAI's coding agent platform supporting parallel agents, background automations, scheduled tasks, and configurable agent personalities. Powered by GPT-5.2-Codex.
One-Shot App Generation
The ability to generate a complete, working application from a single prompt, including UI, backend logic, and deployment.
P
Prompt Engineering
The practice of crafting effective natural language instructions for AI models, including clear prompts that describe desired behavior, constraints, and context.
Parallel Agent Execution
Running multiple agentic coding tasks simultaneously, where each agent operates on its own git branch and merges back when complete.
R
Replit
Browser-based development environment with an Autonomous AI Agent that builds, tests, and deploys applications end-to-end with no local setup required.
RAG (Retrieval-Augmented Generation)
A technique where AI models retrieve relevant information from external sources before generating responses. MCP enables a form of RAG through servers.
S
Semantic Search
Search based on meaning rather than exact keyword matching. Used by MCP servers and vector databases to find relevant code and documentation.
Steering Files
Configuration files in Kiro that define how agents interact with your project or globally, adding context, coding standards, and preferred workflows.
Sam Altman
CEO of OpenAI, the company behind Codex and GPT models.
SWE-bench
A coding benchmark that tests AI's ability to fix real-world software bugs from open source repositories. Used to compare practical coding ability of different AI models.
Semgrep MCP
Security-focused MCP server for real-time static analysis and vulnerability detection as the AI writes code.
Spec-Driven Development
Kiro's approach where natural language requirements are converted into structured specifications with acceptance criteria before code is generated.
Supabase MCP
MCP server for full Supabase database and authentication access from AI tools.
Structured Data / Schema Markup
Machine-readable metadata added to web pages using Schema.org vocabulary that helps search engines and AI systems understand content.
V
Vibe Coding
An approach to software development where you describe what you want in natural language and an AI generates the code. Coined by Andrej Karpathy in February 2025.
v0 by Vercel
Specialized tool for generating React/Tailwind UI components from text or design descriptions, producing clean code using shadcn/ui.
VibeOps
An emerging term for the operational practices around deploying, monitoring, and maintaining vibe-coded applications in production.
Vibe Working
A term coined by Scott White of Anthropic in February 2026 to describe the expansion of AI-assisted workflows beyond coding into all knowledge work.
W
Stay in the loop
Get weekly updates on trending AI coding tools and projects.