Glossary
This glossary provides clear definitions of technical terms, concepts, and jargon used throughout AI-assisted development with Cursor IDE, Claude Code, and MCP servers.
Agent Mode
Section titled “Agent Mode”A Cursor feature where AI can autonomously read, understand, and modify code across multiple files. Unlike Ask mode, Agent mode can make actual changes to your codebase and execute complex multi-step tasks.
Agentic AI
Section titled “Agentic AI”AI systems capable of autonomous, goal-directed behavior with the ability to plan, execute, and iterate on tasks without constant human guidance. Both Claude Code and Cursor’s Agent mode exemplify agentic AI.
API Key
Section titled “API Key”A unique identifier used to authenticate requests to AI services. Can be personal (tied to individual accounts) or organizational (shared across teams).
Ask Mode
Section titled “Ask Mode”Cursor’s read-only chat mode for exploring and understanding code without making changes. Ideal for Q&A, debugging discussions, and getting suggestions before implementation.
@ Symbols
Section titled “@ Symbols”Context references in Cursor that allow you to include specific files, folders, documentation, or other resources in your AI conversation. Examples: @file.js, @docs, @codebase.
Background Agent
Section titled “Background Agent”Cursor’s asynchronous task runner that executes long-running operations while you continue working. Can be triggered via editor or external integrations like Slack.
BugBot
Section titled “BugBot”Cursor’s automated code review feature that analyzes pull requests for potential bugs, security issues, and code quality problems. Integrates with GitHub for seamless PR reviews.
Checkpoint
Section titled “Checkpoint”A saved state in Cursor’s chat or composer session that you can return to if needed. Useful for experimenting with different approaches without losing progress.
CLAUDE.md
Section titled “CLAUDE.md”A special markdown file that Claude Code automatically reads for project-specific context, including coding standards, common commands, and project structure. Can be placed at project root or in subdirectories.
CLI (Command Line Interface)
Section titled “CLI (Command Line Interface)”Text-based interface for interacting with programs. Claude Code operates primarily through CLI, while Cursor provides a graphical interface.
Code Context
Section titled “Code Context”The surrounding code, files, and project information that AI uses to understand your request and generate appropriate responses. Includes imports, function definitions, and related files.
Codebase Indexing
Section titled “Codebase Indexing”The process of analyzing and creating searchable representations of your entire project. Cursor uses embeddings for semantic search, while Claude Code uses agentic search.
Composer
Section titled “Composer”Cursor’s multi-file editing mode that allows you to describe changes affecting multiple files and see all modifications in a unified interface before applying them.
Context Window
Section titled “Context Window”The maximum amount of text (measured in tokens) that an AI model can process in a single request. Varies by model: Claude 4 Sonnet (200k tokens), GPT-4 (128k tokens).
Deep Reasoning
Section titled “Deep Reasoning”Claude Code’s enhanced thinking modes (/think, /ultrathink) that allocate more computational resources for complex problem-solving and architectural decisions.
Diff View
Section titled “Diff View”Visual representation showing the differences between original code and AI-suggested changes. Both tools provide inline diff views for easy review.
Embeddings
Section titled “Embeddings”Mathematical representations of code that enable semantic search and understanding. Cursor creates embeddings of your codebase for intelligent context retrieval.
Enterprise Mode
Section titled “Enterprise Mode”Features and configurations designed for corporate environments, including privacy controls, SSO integration, and compliance certifications (SOC 2, GDPR).
Fast Requests
Section titled “Fast Requests”Priority API calls in subscription plans that process quickly without rate limiting. Limited by plan tier (e.g., 500/month for Cursor Pro).
Fusion Model
Section titled “Fusion Model”Cursor’s advanced Tab completion model capable of handling long-range code jumps and maintaining context across larger code sections.
Git Integration
Section titled “Git Integration”Built-in version control features allowing AI to read commit history, create branches, write commit messages, and manage pull requests.
Claude Code’s automation system for triggering actions at specific points in the development workflow. Includes pre/post command hooks and custom event handlers.
IDE (Integrated Development Environment)
Section titled “IDE (Integrated Development Environment)”A comprehensive software application for coding. Cursor is an AI-enhanced IDE, while Claude Code integrates with existing IDEs via extensions.
Inline Edit
Section titled “Inline Edit”Cursor’s Ctrl/Cmd+K feature for making AI-powered edits to selected code directly in the editor without opening a separate chat interface.
Intent Context
Section titled “Intent Context”The part of an AI prompt that describes what you want to achieve (prescriptive), as opposed to state context which describes the current situation (descriptive).
Large Language Model (LLM)
Section titled “Large Language Model (LLM)”AI systems trained on vast text datasets to understand and generate human-like text. Examples: Claude 4 Opus, GPT-4, Gemini 2.5 Pro.
Long Context Mode
Section titled “Long Context Mode”Special mode for handling larger amounts of code/text that exceeds normal context limits. Consumes additional credits but enables processing of entire large files.
Max Mode
Section titled “Max Mode”Enhanced context mode in Cursor that uses token-based pricing to provide maximum model capabilities and context window for complex tasks.
MCP (Model Context Protocol)
Section titled “MCP (Model Context Protocol)”Open standard for connecting AI assistants to external tools and data sources. Enables integrations with databases, APIs, browsers, and custom tools.
Memory System
Section titled “Memory System”Persistent storage of project-specific information. Cursor uses “Memories” feature, while Claude Code uses CLAUDE.md files and the /memory command.
Message Limit
Section titled “Message Limit”Rate limiting mechanism based on number of interactions within a time window. Claude Code Pro allows ~45 messages per 5-hour session.
Multi-file Editing
Section titled “Multi-file Editing”The ability to make coordinated changes across multiple files in a single operation. Both tools excel at this through different interfaces.
Opus 4
Section titled “Opus 4”Anthropic’s most powerful model (5x cost of Sonnet 4), best for complex architectural planning, system design, and deep code analysis.
PAT (Personal Access Token)
Section titled “PAT (Personal Access Token)”Secure authentication method for Git services like GitHub. Required for CLI operations and certain MCP server integrations.
Planning Mode
Section titled “Planning Mode”Explicit mode in both tools for breaking down complex tasks into structured steps before implementation. Triggered by keywords like “plan” or “think”.
PRD (Product Requirements Document)
Section titled “PRD (Product Requirements Document)”Detailed specification of features and functionality. Both tools can generate, interpret, and implement code based on PRDs.
Privacy Mode
Section titled “Privacy Mode”Security feature that prevents code and data from being used for model training. Essential for enterprise compliance and IP protection.
Prompt Engineering
Section titled “Prompt Engineering”The practice of crafting effective instructions for AI models to get optimal results. Includes techniques like few-shot examples and chain-of-thought reasoning.
Rules Directory
Section titled “Rules Directory”Cursor’s .cursor/rules/ folder containing project-specific AI behavior guidelines. Automatically loaded based on context.
Semantic Search
Section titled “Semantic Search”AI-powered search that understands meaning and intent rather than just matching keywords. Used for finding relevant code across large codebases.
Slash Commands
Section titled “Slash Commands”Special commands prefixed with / that trigger specific actions. Examples: /help, /clear, /memory, /think.
Slow Requests
Section titled “Slow Requests”Rate-limited API calls in free tiers that process with delays. Unlimited in quantity but increasingly slow under heavy use.
Security compliance certification ensuring proper handling of customer data. Both Cursor and Claude Code (Enterprise) maintain SOC 2 compliance.
Sonnet 4
Section titled “Sonnet 4”Anthropic’s balanced model offering excellent performance at reasonable cost. Default model for most coding tasks in both tools.
SSO (Single Sign-On)
Section titled “SSO (Single Sign-On)”Enterprise authentication allowing users to access multiple applications with one set of credentials.
State Context
Section titled “State Context”Information describing the current state of code, errors, or environment (descriptive), as opposed to intent context which describes desired outcomes.
Subagent
Section titled “Subagent”In Claude Code, a separate AI instance spawned to handle specific subtasks while preserving the main conversation’s context.
Tab Autocomplete
Section titled “Tab Autocomplete”Cursor’s predictive text feature that suggests multi-line code completions based on context. Accepts suggestions with Tab key.
Terminal Integration
Section titled “Terminal Integration”Direct access to command line from within the tool. Claude Code is terminal-native, while Cursor provides integrated terminal panels.
Thinking Tokens
Section titled “Thinking Tokens”Computational budget allocated for AI reasoning. Higher-tier models and “think” modes use more thinking tokens for better results.
Todo List Approach
Section titled “Todo List Approach”Workflow pattern where complex tasks are broken into checkboxes that AI works through systematically, updating progress as it goes.
Basic unit of text processed by LLMs, typically 3-4 characters. Context windows and pricing are measured in tokens.
Ultrathink
Section titled “Ultrathink”Claude Code’s maximum reasoning mode, allocating the most computational resources for extremely complex problems requiring deep analysis.
Version Control Integration
Section titled “Version Control Integration”Built-in support for Git operations including commits, branches, and pull requests without leaving the development environment.
Workspace
Section titled “Workspace”The root directory and all subdirectories of your current project. Both tools can analyze and understand entire workspace structures.
Yolo Mode
Section titled “Yolo Mode”High-trust mode where AI can execute commands and make changes without requesting permission for each action. Use with caution.
Quick Navigation
Section titled “Quick Navigation”Core Concepts: Agent Mode | Context Window | LLM | MCP | Token
Cursor-Specific: Ask Mode | Background Agent | BugBot | Composer | Max Mode
Claude Code-Specific: CLAUDE.md | Deep Reasoning | Hooks | Subagent | Ultrathink
Security & Enterprise: Privacy Mode | SOC 2 | SSO | Enterprise Mode
Workflow Patterns: Planning Mode | Todo List Approach | Multi-file Editing