Claude Code
terminal CLIVS CodeDesktopwebJetBrains
Claude Code has the broadest explicit extension taxonomy in a single .claude ecosystem. CLAUDE.md supports managed policy, project, user, and ancestor directory layers with @import for recursive file references including AGENTS.md. Hooks are the most capable with 4 handler types across 26 lifecycle events, and Agent Teams (experimental) enable multi-session coordination with shared tasks and peer-to-peer messaging.
CLAUDE.md and Rules
InstructionsVendor Terms CLAUDE.md, CLAUDE.local.md, rules, @import syntax, auto memory, AGENTS.md
Interfaces terminal CLI, VS Code, Desktop, web, JetBrains
Scopes
Organization / EnterpriseUser / HomeProject / Repo RootSubdirectory / Folder
Availability current (explicit)
Trust Model Contextual guidance, not enforced configuration; compliance not guaranteed for vague or conflicting instructions
Notes Walks up directory tree with on-demand subdirectory loading; rules can be path-scoped; @import syntax with 5-hop recursive depth supports AGENTS.md; managed policy CLAUDE.md at OS-level paths cannot be excluded; auto memory is a complementary per-working-tree system
Skills
PromptsSkills
SkillsVendor Terms skills, SKILL.md, commands, slash commands, Agent Skills standard
Interfaces terminal CLI, VS Code, Desktop, web, JetBrains
Scopes
Project / Repo RootUser / HomeOrganization / Enterprise
Availability current (explicit)
Trust Model Reusable knowledge and workflows loaded on demand or invoked via /name; enterprise deployment via managed settings
Notes Reference skills provide knowledge, action skills trigger workflows; follows Agent Skills open standard (agentskills.io); many built-in skills (/batch, /debug, /simplify, /ultraplan, /schedule); enterprise deployment via managed settings
MCP Servers
MCP & ToolsVendor Terms MCP servers, .mcp.json, managed-mcp.json, MCP Tool Search
Interfaces terminal CLI, VS Code, Desktop, web, JetBrains
Scopes
User / HomeProject / Repo RootOrganization / Enterprise
Availability current (explicit)
Trust Model Project-scoped servers require explicit approval; enterprise lockdown via managed-mcp.json allowlist/denylist
Notes Five-tier scope hierarchy (local > project > user > plugin > claude.ai connectors); Tool Search defers tool loading until needed; supports HTTP, SSE, and stdio transports; OAuth 2.0 with scopes pinning; Claude Code can serve as an MCP server
Subagents
AgentsVendor Terms subagents, custom subagents, managed subagents, agent teams
Interfaces terminal CLI, VS Code, Desktop, web
Scopes
User / HomeProject / Repo RootOrganization / EnterpriseCloud / Web Session
Availability current (explicit)
Trust Model Isolated workers with their own context, prompt, and tools; only summary returns to parent
Notes Five priority levels (managed > CLI flag > project > user > plugin); built-in subagents include Explore, Plan, and Guide; supports isolation worktrees, persistent memory, and background mode; managed deployment via managed settings
Agent Teams
AgentsVendor Terms agent teams, teammates, shared task list, peer-to-peer messaging
Interfaces terminal CLI
Scopes
Project / Repo RootUser / Home
Availability experimental (explicit)
Trust Model Teammates inherit lead's permission settings; if lead uses --dangerously-skip-permissions all teammates do too
Notes Multiple independent sessions with shared tasks and peer-to-peer messaging; task dependencies with file locking; plan approval workflow; two display modes (in-process and split-pane)
Hooks
HooksVendor Terms hooks, agent hooks, async hooks
Interfaces terminal CLI, VS Code, Desktop
Scopes
User / HomeProject / Repo RootOrganization / Enterprise
Availability current (explicit)
Trust Model Deterministic scripts that run on lifecycle events; supports shell, HTTP, LLM, and agent handlers; can block tool use
Notes 4 handler types (command, http, prompt, agent) across 26 lifecycle events; async hooks run in background; PreToolUse hooks enforce policy even in bypass mode; if field enables granular tool+argument filtering
Plugins and Marketplaces
Plugins & DistributionVendor Terms plugins, plugin.json, marketplace, .claude-plugin
Interfaces terminal CLI, VS Code, Desktop, web, JetBrains
Scopes
Project / Repo RootUser / HomeOrganization / Enterprise
Availability current (explicit)
Trust Model Packaging layer that executes arbitrary code with user privileges; managed marketplace restrictions allow org lockdown
Notes Includes plugin-only components (LSP servers, monitors) not available standalone; official Anthropic marketplace with submission review; npm packages as plugin source; managed marketplace restrictions for enterprise
Settings and Managed Settings
Settings & PolicyVendor Terms settings.json, managed settings, managed-settings.json, server-managed settings
Interfaces terminal CLI, VS Code, Desktop, web, JetBrains
Scopes
Organization / EnterpriseUser / HomeProject / Repo Root
Availability current (explicit)
Trust Model Admin-delivered managed settings cannot be overridden; controls permissions, hooks, environment variables, and model defaults
Notes Four-tier scope (managed > local > project > user); 60+ settings keys; server-managed settings via Claude.ai admin console; MDM delivery via macOS plist, Windows registry, or file-based; JSON schema at schemastore.org
Output Styles
Settings & PolicyVendor Terms output styles, custom output styles
Interfaces terminal CLI, VS Code, Desktop, web
Scopes
Project / Repo RootUser / Home
Availability current (explicit)
Trust Model Stronger than ordinary context because it changes the system prompt layer; intended to style behavior
Notes Directly modifies the system prompt; three built-in styles (Default, Explanatory, Learning); custom styles as Markdown files; plugins can distribute output styles