Documentation Index

Fetch the complete documentation index at: https://code.claude.com/docs/llms.txt Use this file to discover all available pages before exploring further.

Extend Claude Code

Understand when to use CLAUDE.md, Skills, subagents, hooks, MCP, and plugins.

Claude Code combines a model that reasons about your code with built-in tools for file operations, search, execution, and web access. The built-in tools cover most coding tasks. This guide covers the extension layer: features you add to customize what Claude knows, connect it to external services, and automate workflows.

New to Claude Code? Start with CLAUDE.md for project conventions, then add other extensions as specific triggers come up.

Overview

Extensions plug into different parts of the agentic loop:

• CLAUDE.md adds persistent context Claude sees every session
• Skills add reusable knowledge and invocable workflows
• Code intelligence connects Claude to a language server for symbol-level navigation and live type errors
• MCP connects Claude to external services and tools
• Subagents run their own loops in isolated context, returning summaries
• Dynamic workflows run many subagents from a script Claude writes, returning one result
• Cross-session messaging lets Claude pass a message from one of your sessions to another
• Hooks run your script, HTTP request, MCP tool call, prompt, or subagent when Claude Code reaches a lifecycle event
• Plugins and marketplaces package and distribute these features

Skills are the most flexible extension. A skill is a markdown file containing knowledge, workflows, or instructions. You can invoke skills with a command like /deploy, or Claude can load them automatically when relevant. Skills can run in your current conversation or in an isolated context via subagents.

Match features to your goal

Features range from always-on context that Claude sees every session, to on-demand capabilities you or Claude can invoke, to background automation that runs on specific events. The table below shows what's available and when each one makes sense.

Plugins are the packaging layer. A plugin bundles skills, hooks, subagents, and MCP servers into a single installable unit. Plugin skills are namespaced (like /my-plugin:review) so multiple plugins can coexist. Use plugins when you want to reuse the same setup across multiple repositories or distribute to others via a marketplace.

Build your setup over time

You don't need to configure everything up front. Each feature has a recognizable trigger, and most teams add them in roughly this order:

The same triggers tell you when to update what you already have. A repeated mistake or a recurring review comment is a CLAUDE.md edit, not a one-off correction in chat. A workflow you keep tweaking by hand is a skill that needs another revision.

Compare similar features

Some features can seem similar. For a deeper walkthrough of choosing between them, see Steering Claude Code: when to use CLAUDE.md, skills, hooks, and subagents on the blog. Here's how to tell them apart.

* **Skills** are reusable content you can load into any context
* **Subagents** are isolated workers that run separately from your main conversation

| Aspect                                          | Skill                                          | Subagent                                                         |
| ----------------------------------------------- | ---------------------------------------------- | ---------------------------------------------------------------- |
| **What it is**                                  | Reusable instructions, knowledge, or workflows | Isolated worker with its own context                             |
| **Key benefit**                                 | Share content across contexts                  | Context isolation. Work happens separately, only summary returns |
| **[Context window](/docs/en/context-window) impact** | Adds to your main window                       | Uses a separate window with its own input and output tokens      |
| **Best for**                                    | Reference material, invocable workflows        | Tasks that read many files, parallel work, specialized workers   |

**Skills can be reference or action.** Reference skills provide knowledge Claude uses throughout your session (like your API style guide). Action skills tell Claude to do something specific (like `/deploy` that runs your deployment workflow).

**Use a subagent** when you need context isolation or when your context window is getting full. The subagent might read dozens of files or run extensive searches, but your main conversation only receives a summary. Since subagent work doesn't consume your main context, this is also useful when you don't need the intermediate work to remain visible. Custom subagents can have their own instructions and can preload skills.

**They can combine.** A subagent can preload specific skills (`skills:` field). A skill can run in isolated context using `context: fork`. See [Skills](/docs/en/skills) for details.

| Aspect                    | CLAUDE.md                    | Skill                                   |
| ------------------------- | ---------------------------- | --------------------------------------- |
| **Loads**                 | Every session, automatically | On demand                               |
| **Can include files**     | Yes, with `@path` imports    | Yes, with `@path` imports               |
| **Can trigger workflows** | No                           | Yes, with `/<name>`                     |
| **Best for**              | "Always do X" rules          | Reference material, invocable workflows |

**Put it in CLAUDE.md** if Claude should always know it: coding conventions, build commands, project structure, "never do X" rules.

**Put it in a skill** if it's reference material Claude needs sometimes (API docs, style guides) or a workflow you trigger with `/<name>` (deploy, review, release).

**Rule of thumb:** Keep CLAUDE.md under 200 lines. If it's growing, move reference content to skills or split into [`.claude/rules/`](/docs/en/memory#organize-rules-with-claude/rules/) files.

| Aspect       | CLAUDE.md                           | `.claude/rules/`                                   | Skill                                    |
| ------------ | ----------------------------------- | -------------------------------------------------- | ---------------------------------------- |
| **Loads**    | Every session                       | Every session, or when matching files are opened   | On demand, when invoked or relevant      |
| **Scope**    | Whole project                       | Can be scoped to file paths                        | Task-specific                            |
| **Best for** | Core conventions and build commands | Language-specific or directory-specific guidelines | Reference material, repeatable workflows |

**Use CLAUDE.md** for instructions every session needs: build commands, test conventions, project architecture.

**Use rules** to keep CLAUDE.md focused. Rules with [`paths` frontmatter](/docs/en/memory#path-specific-rules) only load when Claude works with matching files, saving context.

**Use skills** for content Claude only needs sometimes, like API documentation or a deployment checklist you trigger with `/<name>`.

* **Subagents** are workers Claude spawns, each returning a summary to the conversation that spawned it
* **[Dynamic workflows](/docs/en/workflows)** are scripts Claude writes that run many subagents in the background and return one result

**Use a subagent** when you need a quick, focused worker: research a question, verify a claim, review a file. The subagent does the work and returns a summary, so your main conversation stays clean. Subagents that Claude named when it spawned them can also [message each other](/docs/en/sub-agents#what-loads-at-startup).

**Use a dynamic workflow** when a job [outgrows a handful of subagents](/docs/en/workflows#when-to-use-a-workflow), or when you want the findings cross-checked before you see them, such as a codebase-wide audit, a large migration, or a plan drafted from several angles. To start one, [ask for a workflow in your prompt](/docs/en/workflows#ask-for-a-workflow-in-your-prompt).

**To pass a finding from one of your sessions to another**, ask the first session's Claude to send it. Claude delivers it with [cross-session messaging](/docs/en/cross-session-messaging). [Run agents in parallel](/docs/en/agents) compares the other ways to run more than one Claude at once, including sessions you hand off and check back on later.

| Aspect         | MCP                                                  | Skill                                                   |
| -------------- | ---------------------------------------------------- | ------------------------------------------------------- |
| **What it is** | Protocol for connecting to external services         | Knowledge, workflows, and reference material            |
| **Provides**   | Tools and data access                                | Knowledge, workflows, reference material                |
| **Examples**   | Slack integration, database queries, browser control | Code review checklist, deploy workflow, API style guide |

These solve different problems and work well together:

**MCP** gives Claude purpose-built tools for an external system, with the connection and authentication handled by the server.

**Skills** give Claude knowledge about how to use those tools effectively, plus workflows you can trigger with `/<name>`. A skill might include your team's database schema and query patterns, or a `/post-to-slack` workflow with your team's message formatting rules.

| Aspect           | Hook                                                                              | Skill                                                                 |
| ---------------- | --------------------------------------------------------------------------------- | --------------------------------------------------------------------- |
| **Runs**         | A shell command, HTTP request, MCP tool call, LLM prompt, or subagent             | Instructions Claude reads and follows                                 |
| **Triggered by** | [Lifecycle events](/docs/en/hooks#hook-events) such as `PostToolUse` or `SessionStart` | You typing `/<name>`, or Claude matching the description to your task |
| **Determinism**  | Always fires on its event; the trigger is guaranteed                              | Claude interprets the instructions; outcome can vary                  |
| **Context cost** | Zero unless the hook returns output                                               | Description loads each session; full content loads when used          |
| **Best for**     | Linting after edits, blocking unsafe commands, logging, notifications             | Workflows that need reasoning, reference material, multi-step tasks   |

**Use a hook** when the action must happen the same way every time and doesn't need Claude to think. For example: format on save, reject `rm -rf /`, post a Slack message when a session ends.

**Use a skill** when Claude should decide how to apply the steps, or when the content is knowledge rather than a script. For example: a `/release` checklist, your API style guide, a debugging playbook.

**Put guardrails in hooks.** An instruction like "never edit `.env`" in CLAUDE.md or a skill is a request, not a guarantee. A `PreToolUse` hook that blocks the edit is enforcement. If a rule must hold every time, make it a hook rather than a prompt instruction.

**Hook output lands in context.** A `PostToolUse` hook that runs your linter feeds results back as text Claude reads; a `/fix-lint` skill tells Claude how to resolve them.

Understand how features layer

Features can be defined at multiple levels: user-wide, per-project, via plugins, or through managed policies. You can also nest CLAUDE.md files in subdirectories or place skills in specific packages of a monorepo. When the same feature exists at multiple levels, here's how they layer:

• CLAUDE.md files are additive: all levels contribute content to Claude's context simultaneously. Files from your working directory and above load at launch; subdirectories load as you work in them. When instructions conflict, Claude uses judgment to reconcile them, with more specific instructions typically taking precedence. See how CLAUDE.md files load.
• Skills and subagents override by name: when the same name exists at multiple levels, one definition wins based on priority (managed > user > project for skills; managed > CLI flag > project > user > plugin for subagents). Plugin skills are namespaced to avoid conflicts. See skill discovery and subagent scope.
• MCP servers override by name: local > project > user. See MCP scope.
• Hooks merge: all registered hooks fire for their matching events regardless of source. See hooks.

Combine features

Each extension solves a different problem: CLAUDE.md handles always-on context, skills handle on-demand knowledge and workflows, MCP handles external connections, subagents handle isolation, and hooks handle automation. Real setups combine them based on your workflow.

For example, you might use CLAUDE.md for project conventions, a skill for your deployment workflow, MCP to connect to your database, and a hook to run linting after every edit. Each feature handles what it's best at.

Understand context costs

Every feature you add consumes some of Claude's context. Too much can fill up your context window, but it can also add noise that makes Claude less effective; skills may not trigger correctly, or Claude may lose track of your conventions. Understanding these trade-offs helps you build an effective setup. For an interactive view of how these features combine in a running session, see Explore the context window.

Context cost by feature

Each feature has a different loading strategy and context cost:

*By default, skill descriptions load at session start so Claude can decide when to use them. Set disable-model-invocation: true in a skill's frontmatter to hide it from Claude entirely until you invoke it manually. For a skill you didn't write, set skillOverrides in settings to do the same without editing its file.

Understand how features load

Each feature loads at different points in your session. The tabs below explain when each one loads and what goes into context.

**What loads:** Full content of all CLAUDE.md files (managed, user, and project levels).

**Inheritance:** Claude reads CLAUDE.md files from your working directory up to the root, and discovers nested ones in subdirectories as it accesses those files. See [How CLAUDE.md files load](/docs/en/memory#how-claude-md-files-load) for details.

<Tip>Keep CLAUDE.md under 200 lines. Move reference material to skills, which load on demand. To get [trim proposals for a checked-in CLAUDE.md](/docs/en/memory#my-claude-md-is-too-large), run `/doctor`.</Tip>

**When:** Depends on the skill's configuration. By default, descriptions load at session start and full content loads when used. For user-only skills (`disable-model-invocation: true`), nothing loads until you invoke them.

**What loads:** For model-invocable skills, Claude sees names and descriptions in every request. When you invoke a skill with `/<name>` or Claude loads it automatically, the full content loads into your conversation.

**How Claude chooses skills:** Claude matches your task against skill descriptions to decide which are relevant. If descriptions are vague or overlap, Claude may load the wrong skill or miss one that would help. To tell Claude to use a specific skill, invoke it with `/<name>`. Skills with `disable-model-invocation: true` are invisible to Claude until you invoke them.

**Context cost:** Low until used. User-only skills have zero cost until invoked.

**In subagents:** Skills work differently in subagents. Instead of on-demand loading, skills listed in the subagent's `skills` field are fully preloaded into its context at launch. Subagents can still discover and invoke unlisted project, user, and plugin skills through the Skill tool.

<Tip>Use `disable-model-invocation: true` for skills with side effects. This saves context and ensures only you trigger them.</Tip>

**What loads:** Tool names and server instructions from connected servers. Full JSON schemas stay deferred until Claude needs a specific tool.

**Context cost:** [Tool search](/docs/en/mcp#scale-with-mcp-tool-search) is on by default, so idle MCP tools consume minimal context.

<Tip>Run `/mcp` to see each server's connection status. Run `/context all` to see how many tokens each loaded MCP tool uses. Claude Code [reconnects to remote servers automatically](/docs/en/mcp#automatic-reconnection) if they drop, and you can disconnect servers you're not actively using.</Tip>

**What loads:** Type errors and warnings after each file edit. Definition, reference, and type information when Claude looks up a symbol.

**Context cost:** Low. Symbol lookups often replace broad file reads, so net context use can go down.

<Tip>The LSP tool is inactive until you install a [code intelligence plugin](/docs/en/discover-plugins#code-intelligence) for your language.</Tip>

**What loads:** Fresh, isolated context containing:

* The agent's own system prompt, not the Claude Code system prompt
* Full content of skills listed in the agent's `skills:` field
* CLAUDE.md and git status, except the built-in Explore and Plan agents [omit both](/docs/en/sub-agents#what-loads-at-startup)
* Whatever context the lead agent passes in the prompt

For a [fork](/docs/en/sub-agents#fork-the-current-conversation), Claude Code loads the parent's conversation so far, system prompt, and tools instead.

**Context cost:** Isolated from main session.

<Tip>Use subagents for work that doesn't need your full conversation context. Their isolation prevents bloating your main session.</Tip>

**What loads:** Nothing by default. Hooks execute outside the main conversation.

**Context cost:** Zero, unless the hook returns output that gets added as messages to your conversation.

<Tip>Hooks are ideal for side effects (linting, logging) that don't need to affect Claude's context.</Tip>

Learn more

Each feature has its own guide with setup instructions, examples, and configuration options.