Claude Code adds shareable mods for its behavior and interface


The extensions work in the command-line and desktop apps and can alter prompts and tool calls. Permission protections differ across environments.
Anthropic has added mods to its Claude Code developer tool, letting users change the app’s behavior and interface—not just the instructions its superintelligence (SI) follows. The extensions are enabled by default in version 2.1.287 and later and are available in both the command-line interface and desktop app, The New Stack reports.
Earlier customization included settings, persistent instructions in CLAUDE.md files and hooks that respond to events. Those controls could shape Claude’s instructions, permissions and available tools, alongside visual tweaks such as status lines and output styles. They could not change Claude Code’s own features or the rest of its interface.
Mods are small JavaScript or TypeScript functions that respond to events inside Claude Code. They can rewrite a prompt before it reaches the SI model, block or rewrite a request to use a tool, handle permission requests, replace interface elements or add functionality.
A mod’s code remains active throughout a session, so it can retain information between events. That allows mods to update displays live, launch processes, add commands or make new tools available to the model. Developers can package mods as plugins—installable extensions—and share them through plugin marketplaces, including GitHub repositories.
In Anthropic’s launch blog, technical staff member Addy Osmani demonstrates Token Weather. The example mod tracks use of the context window—the amount of information the model can work with at once—and displays that usage above the prompt. The New Stack also reports that a developer built a mod for passing credentials to Claude without leaving the underlying secret in conversation history.
Installing a mod means running code locally with access, through Claude Code, to the developer’s machine. Anthropic recommends inspecting the code and installing only mods from trusted publishers.
According to The New Stack, mods inherit Claude Code’s existing plugin controls. For Team and Enterprise users, and on machines with managed settings, Anthropic’s sec-default guard loads before user-installed mods. By default, this safeguard prevents those mods from overriding rules that deny permission. Outside those guarded environments, mods can override some permission decisions.

A security expert says the macOS permission can make chat files readable at the system level. Meta says Muse also needs its own Messages setting enabled.
2 min

The reported increases affect four models at Samsung.com and other retailers. The entry-level A17 is now $70 above its January launch price.
1 min

QSSR is now available in Wolverine and Ghost of Yōtei. Digital Foundry found smaller gains in Ghost of Yōtei and a substantial gap versus PS5 Pro’s PSSR.
3 min