Claude Kit
FREE CLAUDE CODE
TEMPLATES
Skills, subagents, hooks, MCP servers and CLAUDE.md files to copy into a project today. Each says where the file goes and is checked against Anthropic's docs.
Reusable know-how Claude loads when a task matches — and the /slash commands you type yourself. All Skills →
Specialist assistants Claude Code delegates to, each with its own instructions, tools and context window. All Subagents →
Shell commands Claude Code runs at fixed moments — before a tool, after an edit, when it needs you. All Hooks →
Give Claude new tools by writing a Model Context Protocol server, then plugging it into Claude Code. All MCP Servers →
The memory file Claude reads at the start of every session — your project's rules, commands and traps. All CLAUDE.md →
WHICH ONE DO I NEED?
Subagent, skill, hook or MCP server
| Kind | What it is | Who triggers it | Use it for | Lives in |
|---|---|---|---|---|
| Subagent | A separate assistant with its own context, instructions and tools | Claude delegates, or you ask for it by name | Noisy or specialist jobs — reviews, debugging — kept out of your main context | .claude/agents/*.md |
| Skill | Instructions, and optionally scripts, loaded when relevant | Claude when the description matches, or you with /name |
Procedures and conventions — how to commit, how to build an endpoint | .claude/skills/<name>/SKILL.md |
| Hook | A shell command Claude Code itself runs at a fixed moment | Always — an event, not a decision | Rules that must hold every time — formatting, protected files, notifications | "hooks" in settings.json |
| MCP server | A program that gives Claude new tools | Claude, when a tool fits the task | Reaching systems Claude cannot see — databases, APIs, your notes | claude mcp add or .mcp.json |
| CLAUDE.md | Memory loaded at the start of every session | Always loaded | The commands, layout, conventions and traps of one project | ./CLAUDE.md, ~/.claude/CLAUDE.md |
QUESTIONS
Claude Code templates, answered
What is the difference between a subagent and a skill?
A subagent is a separate assistant with its own context window, instructions and tool list — Claude hands it a task and gets a summary back. A skill is instructions (and optionally scripts) loaded into the current conversation when the task matches. Use a subagent to keep a noisy job out of your main context; use a skill to teach Claude a procedure.
When should I use a hook instead of a skill?
When it must happen every time. A skill is guidance Claude chooses to follow; a hook is a command Claude Code runs itself at a fixed moment, such as before every file edit. Formatting, blocking protected files and notifications belong in hooks.
Where do Claude Code subagent files go?
In .claude/agents/ inside a project, or ~/.claude/agents/ for every project on your machine. A project agent overrides a personal one with the same name.
Are custom slash commands still supported?
Yes, but they have been merged into skills. A file at .claude/commands/deploy.md and a skill at .claude/skills/deploy/SKILL.md both create /deploy; Anthropic recommends skills for new work.
Are these templates free to use?
Yes. Copy them, change them, ship them. Each page shows the date its format was last checked and links to Anthropic's documentation, which wins whenever the two disagree.
Calling Claude from your own code rather than from Claude Code? The Anthropic API recipe has the Messages API in C#, Python, Java and JavaScript, and the MCP recipe goes deeper on writing servers.
NEW CODE EVERY MONTH.
Get a note when new Claude Code templates, code recipes or tools go live.
No account required
We email once to confirm. Unsubscribe anytime — it deletes your address.