Anthropic has added TypeScript mods to Claude Code, letting developers change how the coding assistant handles prompts, tool calls and parts of its interface. Mods ship inside plugins and are available in the command-line tool and desktop app, the company said on Oct. 1.
A mod is a function that runs when Claude Code reaches an event, such as receiving a prompt, making a tool call or drawing part of the interface. For example, a mod can change a prompt before the model sees it, block or retry a tool call, approve or deny a permission request, redact tool output, or add panes and buttons.
Anthropic says existing settings hooks can run commands around events, but cannot change prompts or tool calls, add interface elements, or replace built-in features. Claude Code's /diff feature now ships as a mod, so users can turn it off or replace it through /plugin. Anthropic says it plans to move more built-in features into mods.
That flexibility comes with access to the user's machine. Anthropic's documentation says mods run inside Claude Code, are not sandboxed, and have the user's access to files, processes and the network. Because a mod can also change permission decisions, review its source and install only from trusted providers. Team and Enterprise administrators can restrict which plugins and mods load, but those controls do not sandbox a mod that is allowed to run.
Sources
- Anthropic's announcement, Oct. 1, 2026: availability, event handling, examples and
/diff. - Claude Code mod documentation: how handlers work and where they run.
- Organization controls for mods: execution privileges and administrator settings.
- Photograph by Slashme via Wikimedia Commons, released under CC0 1.0; illustrative image, not a Claude Code screenshot.
