Claude Code Features/Extensions

Feature	Description	Key capabilities	When to use it	Example	Resources
CLAUDE.md	Persistent context file	Provides consistent instructions/context across every session.	Project conventions, “always do X” rules	“Use pnpm, not npm. Run tests before committing.”	Docs
Skills	Extensible workflows (`SKILL.md`)	Extends knowledge and workflows; can be invoked manually (`/deploy`) or automatically.	Reusable content, reference docs, repeatable tasks	`/deploy` runs your deployment checklist; API docs skill with endpoint patterns	Docs / AgentSkills
MCP	Model Context Protocol	Connects Claude to external services, tools, and data sources.	External data or actions	Query your database, post to Slack, control a browser	Docs
Hooks	Deterministic scripts	Runs OUTSIDE the main LLM loop as pre-defined, reliable scripts.	Automation that must run on every matching event	Run ESLint after every file edit	Docs
Subagents	Isolated execution loops	Runs own loops in isolated context, returning summaries to the main session.	Context isolation, parallel tasks, specialized workers	Research task that reads many files but returns only key findings	Docs
Agent teams	Coordinate multiple independent Claude Code sessions		Parallel research, new feature development, debugging with competing hypotheses	Spawn reviewers to check security, performance, and tests simultaneously
Plugins	Distribution packages	Bundles skills, hooks, subagents, and MCP servers for reuse across repositories (e.g. Claude in Chrome)			Docs, Marketplace

When to Use a Feature

Trigger	Add
Claude gets a convention or command wrong twice	Add it to CLAUDE.md
You keep typing the same prompt to start a task	Save it as a user-invocable skill
You paste the same playbook or multi-step procedure into chat for the third time	Capture it as a skill
You keep copying data from a browser tab Claude can’t see	Connect that system as an MCP server
A side task floods your conversation with output you won’t reference again	Route it through a subagent
You want something to happen every time without asking	Write a hook
A second repository needs the same setup	Package it as a plugin

Invocation Control: Skills can run in the current session or be offloaded to subagents for modularity
Loading Behavior: Descriptions load at session start to minimize latency, while full content is injected only when the skill is triggered
Portability: Use Plugins when you need to share a specific set of tools or logic across multiple different project