Anthropic lets users rewrite how Claude Code works with “mods”, and loads a security mod first on managed machines

Anthropic released mods for Claude Code, its coding agent, on 1 October. The announcement says: “Today we're introducing mods, small TypeScript functions that change how Claude Code works. A mod can rewrite a prompt, add new UI, replace a built-in feature, or add entirely new functionality.”

Users can write a mod themselves or ask Claude Code to write one. Mods ship inside plugins and work in both the command-line tool and the desktop app.

Opening an agent up this far also opens a route for risky changes, and Anthropic addresses that on the same page. On Team and Enterprise plans, and on any machine with managed settings, a built-in mod called sec-default (“security default”) loads first. Anthropic says it stops mods that users install from doing risky things, such as overriding an organisation's permission deny rules. Admins can load their own mods first instead; the page tells them that if they do, they should add sec-default to their list to keep its restrictions.

The design puts a guard rail at the same layer as the customisation: the first mod loaded decides what later ones may change. How well sec-default holds up against a determined mod is Anthropic's claim, and no outside test was found in the window. Anthropic is private.

• Confirmed Mods are TypeScript functions that can rewrite a prompt, add new UI, replace a built-in feature or add new functionality, in the CLI and the desktop app. Claude blog (Anthropic)
• Claimed On Team and Enterprise plans and any machine with managed settings, the built-in sec-default mod loads first and stops user-installed mods from doing risky things such as overriding permission deny rules. Claude blog (Anthropic)