Claude Code now supports mods. A mod is a small TypeScript function, shipped inside a plugin, that hooks into Claude Code itself. Anthropic calls it the biggest change yet to how you can customise the tool.
What a mod can do
- Hold or rewrite a tool call before it runs
- Draw its own UI
- Add a new slash command
- Replace a built-in feature entirely
You don't have to write it
You describe the mod and Claude writes it. Anthropic's own example started with one prompt:
create a mod that draws my context window as a stacked bar above the prompt, one color per category like /context, toggled with /context-bar
Claude wrote the mod, loaded it into the running session, and /context-bar now toggles a live context bar with a legend. No more typing /context to check usage.
Another launch example hides email addresses as pills in the desktop app, revealing one address only when you hover over it. Handy if you share your screen.
What mods are for
The things too niche for Anthropic to ship to everyone: a piece of UI you wish existed, a shortcut for something you type dozens of times a day, or a built-in you'd like to work differently.
How to start
- Update to v2.1.287 or later. Mods work in the CLI and the desktop app.
- Browse Anthropic's sample mods.
- Turn on the built-in "You should know" mod with
/plugin enable cc-plugin-you-should-know@builtin. A small side agent reads along while Claude works and flags anything either of you might have missed. Its code is a good model for your own. - Built something others would use? Plugins can now be published to the Claude directory through a new developer portal. Point it at your repo, follow the review, and choose when it goes live. It reaches everyone using Claude, not only Claude Code users. Run
claude plugin evalfirst to check it's worth shipping.
