Claude Code Terminology Glossary

New to Claude Code customization? This glossary defines common terms you’ll encounter while exploring agents, skills, instructions, and related concepts in the Awesome Claude Code ecosystem.

Use this page as a quick reference when reading articles in the Learning Hub or browsing the repository.

Core Concepts

Agent

A specialized configuration file (*.agent.md) that defines a Claude Code persona or assistant with specific expertise, tools, and behavior patterns. In products that support delegation, the agent is usually the primary coordinator or main session persona, while subagents handle narrower delegated tasks.

When to use: For recurring workflows that benefit from deep tooling integrations and persistent conversational context.

Learn more: What are Agents, Skills, and Instructions

Subagent

A temporary, task-focused agent launched by another agent or orchestrator. A subagent usually gets a narrower prompt, its own isolated context window, and returns a summary back to the main agent instead of staying in the primary conversation.

When to use: For isolated research, parallel analysis, specialized review passes, or delegated implementation steps.

Learn more: Agents and Subagents

Built-in Tool

A native capability provided by Claude Code without requiring additional configuration or MCP servers. Examples include code search, file editing, terminal command execution, and web search. Built-in tools are always available and don’t require installation.

Related terms: Tools, MCP

Chat Mode

Deprecated terminology - This term is no longer used. Use Agent instead.

Previously, “chat mode” was an alternative term for Agent that described how Claude Code Chat could be transformed into domain-specific assistants. The ecosystem has standardized on “Agent” as the preferred terminology.

See: Agent

Collection

Note: Collections are a concept specific to the Awesome Claude Code repository and are not part of standard Claude Code terminology.

A curated grouping of related skills, instructions, and agents organized around a specific theme or workflow. Collections are defined in YAML files (*.collection.yml) in the collections/ directory and help users discover related customizations together.

Example: The “Awesome Claude Code” collection bundles meta-skills for discovering and generating Claude Code customizations.

Learn more: What are Agents, Skills, and Instructions

Custom Agent

See Agent. The term “custom” emphasizes that these are user-defined configurations rather than Claude Code’s default behavior. Custom agents can be created by anyone and shared via repositories like Awesome Claude Code.

Custom Instruction

See Instruction. The term “custom” emphasizes that these are user-defined rules rather than Claude Code’s built-in understanding. Custom instructions are particularly useful for codifying team-specific standards and architectural decisions.

Configuration & Metadata

Front Matter

YAML metadata placed at the beginning of Markdown files (between --- delimiters) that provides structured information about the file and controls its behavior. In this repository, front matter typically includes fields like name, description, mode, model, tools, and applyTo.

The front matter is what controls:

Tool access: Which built-in and MCP tools the customization can use
Model selection: Which AI model powers the customization
Scope: Where the customization applies (e.g., applyTo patterns for instructions)

Note: Not all fields are common across all customization types. Refer to the specific documentation for agents, skills, or instructions to see which fields apply to each type.

Example:

---
name: 'React Component Generator'
description: 'Generate modern React components with TypeScript'
mode: 'agent'
tools: ['codebase']
---

Used in: Skills, agents, instructions, and Learning Hub articles.

Handoff

A VS Code custom-agent frontmatter property (handoffs) that defines suggested transitions from one agent to another, often with a pre-filled follow-up prompt. Handoffs are useful for guided workflows such as research -> implementation or planning -> review.

Important: GitHub’s custom agent configuration reference says handoffs are currently ignored for Claude Code cloud agent on GitHub.com, so this concept is not portable across every Claude Code surface.

Learn more: Agents and Subagents, Building Custom Agents

AGENTS.md

An emerging industry standard file format for defining portable AI coding instructions that work across different AI coding tools (Claude Code, Claude, Codex, and others). The AGENTS.md file, typically placed in a repository root or .github/ directory, contains instructions for how AI assistants should interact with your codebase.

Unlike tool-specific customization files (.agent.md, .prompt.md, .instructions.md), AGENTS.md aims to provide a standardized, platform-agnostic way to define AI behavior that can be consumed by multiple tools.

Key characteristics:

Platform-agnostic format for cross-tool compatibility
Typically contains project context, coding standards, and architectural guidelines
Located at repository root or in .github/ directory

Learn more: AGENTS.md Specification

Related terms: Instruction, Front Matter

Instruction

A configuration file (*.instructions.md) that provides persistent background context and coding standards that Claude Code reads whenever working on matching files. Instructions contain style guides, framework-specific hints, and repository rules that help Claude Code align with your engineering practices automatically.

When to use: For long-lived guidance that applies across many sessions, like coding standards or compliance requirements.

Learn more: What are Agents, Skills, and Instructions, Defining Custom Instructions

Skills & Interactions

Persona

The identity, tone, and behavioral characteristics defined for an Agent. A well-crafted persona helps Claude Code respond consistently and appropriately for specific domains or expertise areas.

Example: A “Database Performance Expert” persona might prioritize query optimization and explain concepts using database-specific terminology.

Related terms: Agent

Prompt

Deprecated — Prompts (*.prompt.md) were reusable chat templates that captured specific tasks or workflows, invoked using the / command in Claude Code Chat. Prompts have been superseded by Skills, which offer the same slash-command invocation plus agent discovery, bundled assets, and cross-platform portability.

If you have existing prompts, consider migrating them to skills. See Creating Effective Skills for guidance.

See: Skill

Skill

A self-contained folder containing a SKILL.md file and optional bundled assets (reference documents, templates, scripts) that packages a reusable capability for Claude Code. Skills follow the open Agent Skills specification and can be invoked by users via /command or discovered and invoked by agents automatically.

Key advantages:

Agent discovery: Extended frontmatter lets agents find and invoke skills automatically
Bundled assets: Reference files, templates, and scripts provide richer context
Cross-platform: Portable across coding agent systems via the Agent Skills specification

Example: A /generate-tests skill might include a SKILL.md with testing instructions, a references/test-patterns.md with common patterns, and a templates/test-template.ts starter file.

When to use: For standardizing how Claude Code responds to recurring tasks, especially when bundled resources improve quality.

Learn more: What are Agents, Skills, and Instructions, Creating Effective Skills

Platform & Integration

MCP (Model Context Protocol)

A standardized protocol for connecting AI assistants like Claude Code to external data sources, tools, and services. MCP servers act as bridges, allowing Claude Code to interact with APIs, databases, file systems, and other resources beyond its built-in capabilities.

Example: An MCP server might provide access to your company’s internal documentation, AWS resources, or a specific database system.

Learn more: Model Context Protocol | MCP Specification | Understanding MCP Servers

Related terms: Tools, Built-in Tool

Hook

A shell command or script that runs automatically in response to lifecycle events during a Claude Code agent session. Hooks are stored as JSON files in .github/hooks/ and can trigger on events like session start/end, prompt submission, before/after tool use, and when errors occur. They provide deterministic automation—linting, formatting, governance scanning—that doesn’t depend on the AI remembering to do it.

Example: A postToolUse hook that runs Prettier after the agent edits files, or a preToolUse hook that blocks dangerous shell commands.

When to use: For deterministic automation that must happen reliably, like formatting code, running linters, or auditing prompts for compliance.

Learn more: Automating with Hooks

Related terms: Agent, Coding Agent

Coding Agent

The autonomous Claude Code agent that works on issues in a cloud environment without continuous human guidance. You assign an issue to Claude Code, it spins up a dev environment, implements a solution, runs tests, and opens a pull request for review.

Key characteristics:

Runs in an isolated cloud environment
Uses your repository’s instructions, agents, skills, and hooks
Always produces a PR—it can’t merge or deploy
Supports iteration via PR comments

When to use: For well-defined tasks with clear acceptance criteria that can be completed autonomously.

Learn more: Using the Claude Code Coding Agent

Related terms: Agent, Hook

Plugin

An installable package that extends Claude Code CLI with a bundled set of agents, skills, hooks, MCP server configurations, and LSP integrations. Plugins provide a way to distribute and share custom capabilities across projects and teams, with versioning, discovery, and one-command installation via marketplaces.

Example: Installing database-data-management@awesome-copilot to get a database specialist agent, migration skills, and schema validation hooks in a single command.

When to use: When you want to share a curated set of Claude Code capabilities across multiple projects or team members, or when you want to install community-contributed tooling without manually copying files.

Learn more: Installing and Using Plugins

Related terms: Agent, Skill, Hook

Tools

Capabilities that Claude Code can invoke to perform actions or retrieve information. Tools fall into two categories:

Built-in tools: Native capabilities like codebase (code search), terminalCommand (running commands), and web (web search)
MCP tools: External integrations provided by MCP servers (e.g., database queries, cloud resource management, or API calls)

Agents and skills can specify which tools they require or recommend in their front matter.

Example front matter:

tools: ['codebase', 'terminalCommand', 'github']

Related terms: MCP, Built-in Tool, Agent

Have a term you’d like to see added? Contributions are welcome! See our Contributing Guidelines for how to suggest additions to this glossary.