Claude Code users now have the ability to request that the tool redesign itself through mods, small plugin-based functions that can rewrite prompts, prevent tool calls, and alter the interface itself.
Previously, developers could customize Claude Code through settings, persistent instructions in CLAUDE.md, hooks, MCP servers, and interface tweaks like status lines and output styles. These mechanisms shaped the instructions Claude follows, its permissions, and available tools, but could not modify Claude Code's core features or interface elements.
Anthropic is now offering substantially greater control through mods, which are small JavaScript or TypeScript functions that integrate with events occurring within Claude Code. These mods can intercept and rewrite a prompt before it reaches the model, block or modify tool calls, handle permission requests, replace sections of Claude Code's interface, or introduce entirely new capabilities.
Each person works differently, so there's no reason why everyone should have an identical Claude experience.
Boris Cherny, Claude Code creator
Changing how Claude Code looks and behaves
Claude Code creator Boris Cherny announced the mods feature on X on Thursday, describing them as a mechanism for developers to transform Claude's appearance and behavior through straightforward prompts, with the ability to package these modifications as plugins for installation by other users.
https://x.com/bcherny/status/2105756563302723721?ref_src=twsrc%5Etfw
The mods can alter what Claude Code displays while processing tasks. A mod might, for instance, display real-time metrics such as the quantity of tool calls Claude has executed, elapsed time, and token consumption, then deliver a summary upon completion.
Within hours of the announcement, developers began experimenting with specialized applications. One developer created a mod that enables passing credentials to Claude while keeping the underlying secret out of the conversation history.
How Claude Code mods work
The concept of mods has been under public development for at least a month. On September 3, Anthropic introduced the idea on GitHub under the technical designation "function hooks," requesting developer input on an API that would permit JavaScript and TypeScript functions to intercept Claude Code events. Six days later, the company announced plans to release the feature imminently under the new name "Claude Mods," while maintaining function hooks as the underlying mechanism.
Mods are now enabled by default in Claude Code 2.1.287 and later versions. Since a mod's code remains active throughout a session, it can retain information between events and respond to Claude Code activity in real time. This capability enables the creation of interface elements that refresh dynamically, initiate processes, add commands, or provide new tools to the model. Anthropic indicates it is already using mods for features such as AGENTS.md support and the /diff pane, with their source code published in the Claude Code repository.
That makes mods a way to fit Claude Code to how you work.
Addy Osmani, member of technical staff at Anthropic
In a technical blog post accompanying the launch, Addy Osmani, a member of technical staff at Anthropic, notes that developers already had multiple options for customizing Claude Code. He explains that mods grant control over Claude Code's own behavior and interface through modules that respond to session events and can modify or replace Claude Code's actions.
Osmani provides an example of an approximately 80-line mod called Token Weather, which monitors context window consumption and displays the result in a band positioned above the prompt. As Claude processes more files, the display transitions from "Clear" at 18% of a 200,000-token window, to "Showers" at 67%, and then "Storm" at 81%, while also indicating recent usage and tokens added in the latest turn.
Since mods are distributed as plugins, developers can publish them through the existing plugin mechanism, including from GitHub repositories. This introduces a security consideration: mod code executes locally with access to the developer's machine through Claude Code, so Anthropic recommends reviewing the source and installing only code from trusted sources.
For organizational use, mods inherit Claude Code's existing plugin controls. Team and Enterprise users and machines with managed settings load Anthropic's security-default guard before user-installed mods. This default configuration prevents those mods from overriding permission-deny rules; in non-guarded environments, mods can override certain permission decisions.
Mods are currently available in the Claude Code CLI and desktop app, and can be installed or shared through plugin marketplaces, including GitHub repositories.