Rewrite Claude Code From the Inside
Mods are Anthropic's new Claude Code feature: small TypeScript or JavaScript functions that hook into Claude Code's events to change how it behaves and how it looks. Most people haven't worked out what to do with them yet. The useful part is that you don't have to write one yourself — tell Claude "build me a mod that does X" and it codes the mod for you.
What a Mod Actually Is
A mod fires on Claude Code's events, which lets it rewrite a prompt before it reaches the model, intercept or block a tool call, replace or add UI, or add new commands. It changes the harness around the model, not the model itself. Mods ship as plugins and install with /plugin.
Three Uses Nobody's Talking About
- Agents that talk to each other. One session plans, another builds, and they pass work between them.
- A live dashboard. Watch your sub-agents, or your "second brain," right inside the terminal.
- Automatic model routing. Send the easy stuff, like typo fixes, to Haiku and keep Opus for hard tasks — so the expensive model's rate limits go to the work that needs it.
One Warning
Mods run with full access to your machine and aren't sandboxed. Only install mods from sources you trust — and read the ones you didn't write.
What's in the Guide
What a mod can hook into, the three uses in more detail, how to have Claude build a mod for you, the safety caveat, and a link to Anthropic's announcement.
The full field guide is in the PDF.
Get the Guide
Drop your email below and we'll send it straight to your inbox.