Claude Mods

"Claude mods" is the community's name for anything that changes how Claude Code behaves. There is no single mod format: Claude Code has several official extension points, each stored as plain files in your project's .claude/ folder or your user settings. Project rules live in CLAUDE.md. Reusable prompts become slash commands or skills. Specialist helpers become subagents. Automatic actions on events become hooks. Outside tools come in through MCP servers. A plugin bundles several of these so a team can install them in one step. This page helps you choose the smallest one that does the job and writes the first draft for you. It is an independent tool, not affiliated with Anthropic, and it never installs anything on your machine.

CLAUDE.md – a Markdown file Claude reads at the start of every session: coding style, test commands, things to avoid.

Slash commands – a Markdown prompt saved in .claude/commands/ that you run as /name.

Skills – a folder with a SKILL.md file; Claude loads it when a task matches its description, so you don't have to type a command.

Subagents – helpers defined in .claude/agents/ with their own instructions and tool list, e.g. a code reviewer.

Hooks – shell commands in settings.json that run on events such as before or after a tool call.

MCP servers – connect Claude Code to databases, issue trackers or browsers.

Plugins – a bundle of the above, installed from a marketplace with /plugin.

Start with the lightest option. If you just want Claude to follow a rule, write it in CLAUDE.md. If you repeat the same prompt, make a slash command. If Claude should apply a procedure on its own when a task fits, make a skill. If a job needs a separate context and limited tools, make a subagent. If something must happen every time without Claude deciding – formatting, blocking a dangerous command, a notification – use a hook, because hooks are deterministic. Reach for MCP when Claude needs data it cannot get from files, and a plugin when you want to share a set of these with a team.

Popular community mods include a PostToolUse hook that runs Prettier or the project's linter after every edit, a PreToolUse hook that refuses rm -rf or pushes to main, a Stop hook that sends a desktop notification when Claude finishes, a /review slash command for pull requests, and a custom status line that shows the model, git branch and context usage. Type one of these into the generator to get a starter you can adapt.

Mods run with your permissions. A hook executes shell commands automatically, and an MCP server or plugin can read files and reach the network. Before installing a community mod, read its files, check what it runs and where it sends data, prefer well-maintained repositories, and test in a project without secrets. Keep mods in version control so you can see and undo every change.