Anthropic has given developers a way to rewrite how Claude Code behaves. On 1 October it launched mods, small functions written in TypeScript that can change a prompt, add new parts to the interface, replace a built in feature or add something entirely new, according to Anthropic’s announcement. They work in the Claude Code command line tool and in the desktop app.
You do not have to write them yourself. Ask Claude Code for a mod and it can write the code, install it and reload it in the same session. Mods ship inside plugins, so they are installed and shared the same way. If you are new to the tool, our guide explains what Claude Code is and how it differs from Claude.
How mods change Claude Code
Every time Claude Code does something, such as calling a tool, asking for permission or drawing part of the screen, it emits an event. A mod hooks into one of those events and can run before it, after it, instead of it, or around it. With a single function, a mod can block, rewrite or retry a tool call, approve or deny a permission request, or strip secrets from a tool’s output before Claude reads it.
Mods can change what you see too. They can edit or replace parts of the interface, such as a tool result or a question from Claude, and add buttons and inputs that other mods respond to. When several mods hook the same event they run in load order, so mods from different authors can be stacked.
Why Anthropic built them
Anthropic says developers wanted more control without waiting for new features to ship. Hooks, the earlier way to customise Claude Code, could not rewrite events, draw new interface elements or replace features. Anthropic published the design on GitHub before launch to collect feedback.
Some built in features already ship as mods. The diff view is now one, so it can be switched off or replaced with your own version. Anthropic plans to move more features across over time, so teams can strip Claude Code back to a small core and add only what they want.

The catch: mods are not sandboxed
Mods run with the same access to your machine as Claude Code itself. Anthropic says plainly that they are not sandboxed and should only be installed from sources you trust, the same as any other code on your computer. A badly written or malicious mod can do anything Claude Code can.
For organisations, the existing plugin controls apply. Admins can allow or block plugin marketplaces, set in the admin console on Team and Enterprise plans or pushed as managed settings on API plans. On those setups a built in security mod loads first and stops user installed mods from doing risky things, such as overriding rules that deny a permission.
Why Australian businesses should care
For Australian development teams, mods turn company rules into code that runs inside the tool. Anthropic’s own examples include a mod that asks for confirmation before any command touches production configuration, and an audit log that records every call other mods make. Together with secret redaction, that suits teams with obligations under the Privacy Act 1988 or APRA’s CPS 234 information security standard.
It also raises a governance question. Someone has to decide which marketplaces are allowed and who reviews a mod before it spreads across a team. That decision belongs in an AI strategy rather than being left to each developer.
A more open Claude Code
Mods are available now from the Claude directory and through the plugin menu in the command line tool. They continue a run of changes that make Claude Code easier to fit around existing ways of working, after a recent update made switching between Codex and Claude Code easier. The trade off is simple: more power, and more responsibility for what gets installed.
