“No reason why everyone should have an identical Claude experience”: Anthropic’s mods let you change Claude Code’s look and behavior

Anthropic has introduced a new feature called mods for Claude Code, allowing developers to customize the AI's appearance and behavior. These mods are small JavaScript or TypeScript functions that can hook into events within Claude Code, enabling developers to rewrite prompts, block or rewrite tool calls, handle permissions, modify the interface, or add new functionality.

Anthropic's Claude Code creator, Boris Cherny, explains that each user has unique preferences, so there's no need for everyone to have identical Claude experiences. Mods can change what Claude displays while working, such as showing live information like the number of tool calls, runtime, and token consumption, and presenting a summary upon completion.

Users have already started experimenting with more specialized uses of mods, such as passing credentials securely without leaving them in the conversation history. Mods have been in development for about a month, with Anthropic initially sharing the idea on GitHub under the name "function hooks." The feature is now enabled by default in Claude Code 2.1.287 and later.

Anthropic has already integrated mods for features like AGENTS.md support and the /diff pane, and has published their source code for the mods in the Claude Code repository. Developers can publish their mods as plugins through GitHub or other repositories, but should exercise caution as mod code runs locally with access to the developer's machine.

Teams can use mods while still inheriting Claude Code's existing permission controls, and Anthropic recommends installing mods only from trusted sources.