Claude Code: Mods Let Plugins Rewrite Prompts, UI and Built-in Features
Claude CodeAnthropic launched Claude Mods on October 1, 2026, letting plugins ship small TypeScript functions that hook into Claude Code events. A mod can rewrite prompts, block or retry tool calls, answer permission requests, redact tool output and replace interface elements. Some built-in features, such as /diff, now ship as mods, and a first built-in mod called You should know runs a side agent that flags things the user or Claude might miss.
Key Takeaways
- Mods are TypeScript functions shipped inside plugins that hook into Claude Code events and can run before, after or instead of them.
- A mod can rewrite prompts, block or retry tool calls, answer permission requests and redact tool output, which goes well beyond what shell-script hooks allowed.
- Built-in features such as /diff now ship as mods, so users can disable or replace them and slim Claude Code down to a small core.
- Version 2.1.287 adds You should know, a built-in mod with a side agent that flags things the user or Claude might miss, enabled with a /plugin command.
- Mods are not sandboxed and run with the same machine access as Claude Code, so installing one is a real trust decision.
- On Team and Enterprise plans, a
sec-defaultmod loads first to stop user-installed mods from bypassing permission denials.
Sources & Mentions
4 external resources covering this update
What Mods Are
Claude Code now supports mods: small TypeScript functions, packaged inside plugins, that hook into the events Claude Code emits while it works. Anthropic describes the idea simply: a mod can rewrite a prompt, add new UI, replace a built-in feature, or add entirely new functionality. Mods launched on October 1, 2026 in both the Claude Code CLI and the desktop app, and arrived alongside version 2.1.287 with the release note that plugins may now modify deeper behavior.
Until now, plugins mostly supplied instructions, commands or external tools, and hooks were shell scripts that reacted after the fact. Mods run inside Claude Code itself, so they can act before, after or instead of an event, or wrap it with code on both sides.
What a Mod Can Change
Every time Claude Code calls a tool, asks for permission or renders part of its interface, a mod can step in. According to Anthropic, mods can rewrite a prompt before it reaches the model, block, rewrite or retry a tool call, approve or deny a permission request, redact sensitive information from tool output, edit or replace interface elements, and add buttons and interactive inputs. When several mods target the same event, they run in load order, so mods from different authors can stack.
Built-in Mods
Some Claude Code features now ship as mods instead of core code. The /diff feature is the example Anthropic gives: it can be disabled or replaced through plugin settings. Anthropic frames this as a way to pare Claude Code down to a small core and add back only what a user wants.
Version 2.1.287 also adds You should know, a built-in mod in which a side agent watches the session and flags things the user or Claude might miss. It is turned on with /plugin enable cc-plugin-you-should-know@builtin and is available for first-party sessions with telemetry on.
Installing and Writing Mods
Mods ship within plugins, so installation uses the existing /plugin flow and the Claude plugin directory. Developers can submit their own mods by packaging them as plugins. Claude Code can also write a mod on request: a user describes the behavior, Claude generates the TypeScript, installs it and hot reloads it in the same session.
Trust and Team Controls
Mods run with the same machine access as Claude Code and are not sandboxed, so installing one is a trust decision. For Team and Enterprise plans, a built-in security mod named sec-default loads first on managed machines so that user-installed mods cannot get around permission denials. Anthropic suggests team mods for CI/CD status displays, confirmation steps before production config changes, and audit logging.