Skip to content
Founding Lifetime: every Pro skillfirst 100 members · $169 onceUpgrade
Lookbk
Sign In

Claude Code templates: what’s in the big catalog and how to use it

What the claude-code-templates catalog holds, the install command for each type, what each is in Claude Code, and the checks to make before you run it.

For
Claude Code users looking for ready-made skills, agents, hooks and settings
Written for
Claude Code
Time
7 min read · updated Nov 21, 2026

The short answer

“Claude Code templates” usually means the open-source claude-code-templates catalog, whose README says it holds more than 1,980 ready-made components for Claude Code. You install one with npx claude-code-templates@latest --skill <category/name>, and the same command takes --agent, --command, --hook, --mcp and --setting. Each type is a different kind of file in Claude Code, and hooks and MCP servers run code or reach outside services, so preview what a component writes before you install it.

Two things go by “Claude Code templates”. One is a catalog: davila7/claude-code-templates, with its site at aitmpl.com, which collects components you drop into Claude Code. The other is a starting CLAUDE.md for your project. This page covers the catalog, from its README and the npm and GitHub records read on 21 November 2026, and says what each kind of component is in Anthropic’s own Claude Code documentation. The last section is about the CLAUDE.md case.

What the catalog is

The README describes it as an open catalog of components for Claude Code, installable with one command: skills, subagents, slash commands, hooks, MCP servers and settings. It says the CLI writes each file to the right place and merges existing configuration instead of overwriting it. The repository is MIT licensed, was not archived, and had a commit on 21 November 2026. The npm package is at version 1.33.1, MIT, with no preinstall or postinstall script in its registry record.

The catalog by type, with the counts in its README
TypeCountWhat it is in the READMEInstall flag
Skills918Reusable capabilities with progressive disclosure--skill
Agents422Subagents specialized in one domain--agent
Commands288Custom slash commands--command
MCPs119External service integrations--mcp
Settings80Claude Code configuration--setting
Hooks63Automation triggers--hook
Mods50Plugins that change Claude Code’s own interface and behavior--mod
Loops18Autonomous workflows with a goal, an interval and a stop condition--loop
Templates14Complete project configurationssee the catalog
Sandboxes11Launchers for an isolated environment, reference onlynot installed
Plugins1A whole Claude Code plugin from its own marketplace--plugin

The counts add up to 1,984, which is where “more than 1,980” comes from. They change as the catalog does, and the README points to a live counts.json in the repository.

Install one component

Everything runs through npx, so nothing is installed globally. With no flag the CLI opens an interactive browser. A component is named category/name, and the README gives one example per type.

Terminal
npx claude-code-templates@latest --skill creative-design/frontend-design
npx claude-code-templates@latest --agent development-tools/code-reviewer
npx claude-code-templates@latest --command performance/optimize-bundle-size
npx claude-code-templates@latest --mcp database/postgresql-integration
npx claude-code-templates@latest --setting mcp/mcp-timeouts
npx claude-code-templates@latest --hook git/conventional-commits

You can name several in one command, for example an agent, a command and an MCP server together. --search "code review" looks for components from the terminal, and --yes skips the confirmation prompt. The README says every component page on aitmpl.com shows its exact install command.

What each type is in Claude Code

The catalog’s names match the things Claude Code itself defines. Here is each one in Anthropic’s words, with the place Claude Code reads it from.

Component types as Claude Code’s documentation defines them
TypeWhat Claude Code’s docs sayWhere Claude Code reads it
Skill“Skills extend what Claude can do.” A SKILL.md with instructions that Claude uses when relevant, or you call with /skill-name~/.claude/skills/<name>/ for every project, .claude/skills/<name>/ for one
AgentSubagents are specialized AI assistants for specific types of tasks, run in their own context~/.claude/agents/ or .claude/agents/
CommandThe docs say custom commands have been merged into skills, and existing .claude/commands/ files keep working.claude/commands/, or as a skill
HookUser-defined shell commands that Claude Code runs at set points in its lifecycleThe hooks block of a settings file
MCP serverA connection to external tools and data through the Model Context Protocol.mcp.json at the project root for a shared one
SettingJSON keys that change how Claude Code behaves~/.claude/settings.json, .claude/settings.json or .claude/settings.local.json

These locations are from Anthropic’s docs, not from the catalog’s README, which only says its CLI puts files “in the right place”. After an install, look at what changed in your project and your home folder before you rely on it.

Before you run the install command

A skill is instructions. A hook is a shell command that runs on its own, and an MCP server connects Claude Code to a service. Anthropic’s MCP documentation carries a warning to verify you trust each server before connecting it, since servers that fetch outside content can expose you to prompt injection. Those are the types to read most carefully.

Before you install a component from a catalog

The README also says every pull request is checked by Claude Code’s plugin validator and scanned by NVIDIA SkillSpector. That is the project’s own statement, and a scan is not the same as a review of what a component does to your machine. Safety checks before running a skill is the short list for the skill case.

Which to install first

  • Skills are the lowest-risk start: a folder of instructions you can read in a minute. Several of the sources the catalog lists are public skill repositories, which Claude skills on GitHub compares.
  • Settings change permissions and behavior, so read the diff.
  • Hooks and MCP servers last, and only ones whose commands and servers you recognise.

The catalog is one way to find components; Claude Code’s plugin marketplaces are another, and the best Claude Code plugins covers that route. How skills are used and kept tidy is in Claude Code skills best practices.

If you meant a CLAUDE.md template

Anthropic’s memory documentation says CLAUDE.md holds instructions for Claude in every session: ./CLAUDE.md or ./.claude/CLAUDE.md for a project, ~/.claude/CLAUDE.md for all your projects. It says to run /init to generate a starting file: Claude reads your codebase and writes build commands, test instructions and the conventions it finds, and if a file already exists it suggests improvements instead of overwriting it. That fits your own project better than a generic template.

Where Lookbk fits

Lookbk is not a copy of that catalog. It is a small set of skills that make a finished thing, such as a title video, a slide deck or a link-preview card, and each skill page shows the output it rendered before you install it. They install into the same ~/.claude/skills folder as any other skill. These four are free.

Editorial OG Card: an example of what it makes
Free
image
Editorial OG CardClaude CodeCodexCursor+3 more agents
Kinetic Title Drop: an example of what it makes
Free
video
Kinetic Title DropClaude CodeCodexCursor+3 more agents
Swiss Grid Deck: an example of what it makes
Free
slides
Swiss Grid DeckClaude CodeCodexCursor+3 more agents
Mono Editorial Landing: an example of what it makes
Free
web
Mono Editorial LandingClaude CodeCodexCursor+3 more agents

To install one, run the installer with the last part of the skill page’s address, then start a new Claude Code session.

Terminal
curl -fsSL https://lookbk.dev/cli.js | node - add editorial-og-card

Common questions

What is claude-code-templates?
An open-source catalog and command-line installer for Claude Code components. Its README says it holds more than 1,980 skills, agents, commands, hooks, MCP servers and settings, and the repository is MIT licensed.
How do I install a component from it?
Run npx claude-code-templates@latest --skill <category/name>, or use --agent, --command, --hook, --mcp or --setting for the other types. With no flag it opens an interactive browser.
Is it safe to run?
The npm package’s registry record shows no install scripts, but running the CLI writes files into your project, and hooks and MCP servers run code or connect to services. Preview what a component writes, install one at a time and review the changes with git.
Where do the installed files go?
The README only says the CLI puts each file in the right place and merges existing configuration. Claude Code reads skills from .claude/skills, subagents from .claude/agents, hooks and settings from its settings files and project MCP servers from .mcp.json, so check those after an install.
Do I need this to use skills in Claude Code?
No. A skill is a folder with a SKILL.md file that you copy into ~/.claude/skills or .claude/skills, or install from a plugin marketplace.
Can I reuse what is in the catalog?
Check each component’s license. The README says components from other repositories keep their original license and attribution, and lists the sources.
How do I get a CLAUDE.md for my project?
Run /init in Claude Code. Anthropic’s docs say it analyzes your codebase and writes a starting file with build commands, test instructions and conventions, and suggests improvements if one already exists.

Sources

What this guide says about other companies’ tools comes from their own documentation, checked on Nov 21, 2026.

Did this guide help?

See a skill’s result before you install it.

Every skill page shows what it makes, what it needs and which agents can run it.