---
title: "First impressions of Claude Code mods"
description: "Claude Code mods, new in October 2026, enforce rules that hooks and CLAUDE.md only hope Claude remembers. What I built in two days and what worries me."
url: https://www.joekarlsson.com/blog/claude-code-mods-first-impressions/
---

Oct 16, 2026 | 6 min read | [AI](https://www.joekarlsson.com/blog/category/ai/)

# First impressions of Claude Code mods

![A 1921 Ford Model T converted to run on snow, its front wheels swapped for skis and its rear wheels fitted with tracks](https://www.joekarlsson.com/images/blog/claude-code-mods-first-impressions/thumb-d381120f.still.webp)

**Blog** Claude Code mods, new in October 2026, enforce rules that hooks and CLAUDE.md only hope Claude remembers. What I built in two days and what worries me.

**TL;DR** Claude Code mods, new in October 2026, are small programs that plug into Claude Code: they can stop or change what Claude is about to do, and put status lines, side panels and pop-ups on screen. After two days of building them, I think they’re more useful than hooks. Most of my hooks and CLAUDE.md rules only worked if Claude remembered them, and five of my hooks never worked at all without me noticing. A mod enforces the rule itself and shows me it ran. Mods that crash get skipped, they have full access to your machine, and they’re changing fast, so I kept my working blocking hooks as a backup.

On this page

1.  [What Claude Code mods are](https://www.joekarlsson.com/blog/claude-code-mods-first-impressions/#what-claude-code-mods-are)
2.  [Why Claude Code mods beat hooks](https://www.joekarlsson.com/blog/claude-code-mods-first-impressions/#why-claude-code-mods-beat-hooks)
3.  [How I’m using them](https://www.joekarlsson.com/blog/claude-code-mods-first-impressions/#how-im-using-them)
4.  [What I don’t like](https://www.joekarlsson.com/blog/claude-code-mods-first-impressions/#what-i-dont-like)
5.  [Should you use them](https://www.joekarlsson.com/blog/claude-code-mods-first-impressions/#should-you-use-them)

Anthropic shipped [mods](https://code.claude.com/docs/en/plugins/mods/overview) at the start of October 2026. I sat down to build one on October 14, and by the next afternoon I had two Claude Code mods running in every session and more in progress.

**[1](https://www.joekarlsson.com/blog/claude-code-mods-first-impressions/#mn-w-1)** Either the file has grown long enough that the line gets lost, or the line is a rule that can’t have exceptions, in which case it belongs in a hook.[Anthropic's Claude Code best practices](https://code.claude.com/docs/en/best-practices)

Almost every rule I’d given Claude lived in a [CLAUDE.md file](https://www.joekarlsson.com/blog/my-personal-claude-code-skills-repo-accidentally-became-internal-tooling/) Apr 7, 2026**My Personal Claude Code Skills Repo Accidentally Became Internal Tooling** or in one of my Claude Code hooks that handed the model some text and hoped it would remember [1](https://www.joekarlsson.com/blog/claude-code-mods-first-impressions/#mn-1). That’s the honor system.

**[1](https://www.joekarlsson.com/blog/claude-code-mods-first-impressions/#mn-w-1)** Either the file has grown long enough that the line gets lost, or the line is a rule that can’t have exceptions, in which case it belongs in a hook.[Anthropic's Claude Code best practices](https://code.claude.com/docs/en/best-practices)

Two weeks ago, in [my 2026 setup post](https://www.joekarlsson.com/blog/my-development-setup-2026/) Sep 30, 2026**My Development Setup in 2026** , I wrote “If you’re letting an agent near real systems, learn hooks first.” I’d still learn them, but I trust the ones I wrote a lot less now.

## What Claude Code mods are

A mod is a small program that plugs into Claude Code itself. It sees what Claude is about to do, like run a command, edit a file, or answer me, and it can stop it, change it, or let it through. It can also put things on screen: a status line, a side panel, a pop-up, its own slash commands.

Hooks could already stop Claude from doing things. What they couldn’t do was show me anything, or remember anything from one step to the next.

_Diagram: What a Claude Code mod does. Inside Claude Code, Claude is about to run a command, edit a file or answer. A mod sees it first and can stop it, change it, or let it through. The mod can also put things on your screen: a status line, a side panel, a pop-up and its own slash commands. Hooks could already stop Claude; they couldn’t show you anything._

## Why Claude Code mods beat hooks

I was cleaning up my homelab repo and found four hooks that were supposed to hand Claude my notes on Home Assistant, my NAS and Proxmox whenever I mentioned them. None of them had ever worked. One small formatting mistake meant Claude Code threw away everything they sent, and it never said a word about it. The Home Assistant one had been in place since January.

To check, I wrote a test hook that tells Claude a secret word and then asked Claude for the word. With my old format, it had no idea. With the fixed format, it answered right away. In both cases, nothing on my screen told me which one I had.

_Diagram: The same job done by a hook and by a mod: when I mention Home Assistant, hand Claude my notes on it. The hook sent the notes, but one formatting mistake meant Claude Code dropped them, so Claude never saw them, and nothing on my screen said so, broken or fixed. The mod hands Claude the notes and pops up a message on my screen saying it did._

**[2](https://www.joekarlsson.com/blog/claude-code-mods-first-impressions/#mn-w-2)** This was a CLAUDE.md rule first. Edits passed validation, then vanished when Word reopened a stale OneDrive copy.

**[3](https://www.joekarlsson.com/blog/claude-code-mods-first-impressions/#mn-w-3)** Blocking error: your hook exits 2.[Anthropic's hooks docs. Any other failure is non-blocking, and the action goes ahead](https://code.claude.com/docs/en/hooks)

I found a second broken hook while I was at it. Claude and Word documents don’t get along on my Mac [2](https://www.joekarlsson.com/blog/claude-code-mods-first-impressions/#mn-2). Claude would edit a doc in my OneDrive folder, and then Word or OneDrive would save their own copy right over its changes, or the file would come out corrupted. So I wrote a hook to stop Claude from writing into that folder at all. It prints “BLOCKED” and then lets the write go through anyway [3](https://www.joekarlsson.com/blog/claude-code-mods-first-impressions/#mn-3), because it reports the failure the wrong way. As far as I can tell, that hook has never blocked a write.

**[3](https://www.joekarlsson.com/blog/claude-code-mods-first-impressions/#mn-w-3)** Blocking error: your hook exits 2.[Anthropic's hooks docs. Any other failure is non-blocking, and the action goes ahead](https://code.claude.com/docs/en/hooks)

![Always Has Been meme: one astronaut looks at Earth and asks "Wait, my OneDrive guard hook never blocked anything?" while a second astronaut behind him holds a gun and says "Always has been"](https://www.joekarlsson.com/images/blog/claude-code-mods-first-impressions/meme-always-has-been.webp)

Both are my mistakes, and [the docs](https://code.claude.com/docs/en/hooks) explain both. But my problem with hooks, the way I was using them, is that nothing told me when one broke. The rules in my CLAUDE.md file have their own version of that problem. They only work if Claude remembers them in every session. Mods fix the memory problem because the rule is code, and the silent breakage because they can show me they ran.

![Two guys on a bus meme: the sad man by the rock wall is labeled "Rules in CLAUDE.md the agent has to remember" and the happy man looking at the sunny mountains is labeled "Rules with a status line"](https://www.joekarlsson.com/images/blog/claude-code-mods-first-impressions/meme-two-guys-bus.webp)

## How I’m using them

**A global mod** that runs in every project on every machine I use:

![A dark terminal window split into five columns, each a separate Claude Code session with its own colored status line naming the task.](https://www.joekarlsson.com/images/blog/claude-code-mods-first-impressions/five-claude-code-sessions-side-by.webp)

Five Claude Code sessions side by side in Ghostty.

-   **[4](https://www.joekarlsson.com/blog/claude-code-mods-first-impressions/#mn-w-4)**
    
    [My Development Setup in 2026](https://www.joekarlsson.com/blog/my-development-setup-2026/): How I run them: Ghostty splits and git worktrees, with a status line that names each session.
    
    **Parallel sessions.** I usually have several Claude sessions going at once [4](https://www.joekarlsson.com/blog/claude-code-mods-first-impressions/#mn-4). If another one is working in the same folder, the mod stops this one from editing files there or making a commit that could sweep up the other session’s work, and the status line tells me who I’m sharing with.
-   **Word guard.** This is the real fix for the Word mess. If a document is open in Word, Claude can’t write to the file, and the mod tells it to make the change through Word itself instead.
-   **Unpushed work.** If Claude leaves commits I haven’t pushed, a warning shows up under its answer, and I get a notification if I close the session anyway.
-   **Commit messages.** Every commit message has to follow my format, or the commit doesn’t happen.

**A homelab mod** with a side panel that tells me whether anything in my homelab is broken right now. It’s one line, like “1 critical · 8 warnings”, plus the critical items, each with a button that tells Claude to go investigate. It reads the same [health checks and alerts](https://www.joekarlsson.com/blog/homelab-two-years-later/) Mar 5, 2026**My Homelab Two Years Later, Every Upgrade and What It Cost** I already had. The mod also stops Claude from changing my containers by hand and keeps an eye on each deploy until it finishes.

It also replaced the dead hooks. When I mention a service, the mod hands Claude my notes on it and pops up a message saying so. That pop-up is most of why I prefer mods. And unlike my old hooks, every guard has tests that check it says no, which run before I can commit a change to it. If it doesn’t show up, I know right away, instead of nine months later.

**Still in development** is a mod that hands work back to Claude when it skips a step. If Claude changes code without running the tests that cover it, the mod sends it back to run them. It’s a few days old, and I haven’t trusted it with anything that matters yet.

## What I don’t like

**[5](https://www.joekarlsson.com/blog/claude-code-mods-first-impressions/#mn-w-5)** Fail open : when the guard breaks, the action goes through. Fail closed would block everything until the guard works again.

**They fail open.** If a mod crashes [5](https://www.joekarlsson.com/blog/claude-code-mods-first-impressions/#mn-5), Claude Code skips it and keeps going. That makes sense, since one buggy mod shouldn’t freeze every session, but it means a broken guard stops guarding while the session keeps going. My mods load straight from a folder, so when one crashes, Claude Code leaves a one-line note in the chat saying which one. That’s more than my hooks ever gave me. For mods installed from a marketplace, the note only goes to a [debug log](https://code.claude.com/docs/en/plugins/mods/troubleshoot#find-out-why-a-mod-does-nothing). So the visibility argument holds for mods I load myself. I kept my homelab’s old blocking hooks, which do work, running as a backup anyway.

**[6](https://www.joekarlsson.com/blog/claude-code-mods-first-impressions/#mn-w-6)** Mods aren’t sandboxed.[Anthropic's mods overview](https://code.claude.com/docs/en/plugins/mods/overview)

**They can do anything I can.** A mod runs with [6](https://www.joekarlsson.com/blog/claude-code-mods-first-impressions/#mn-6) [full access to my machine](https://code.claude.com/docs/en/plugins/mods/overview). I wrote mine with Claude, so I’m reviewing code I didn’t type before letting it decide which commands run. I wouldn’t install a stranger’s mod without reading every line.

**[6](https://www.joekarlsson.com/blog/claude-code-mods-first-impressions/#mn-w-6)** Mods aren’t sandboxed.[Anthropic's mods overview](https://code.claude.com/docs/en/plugins/mods/overview)

**No screen, no panel.** When Claude runs in the background or in the cloud, mods still do their checks, but nothing shows up, so the status lines are no help there.

**It’s changing fast.** Claude Code put out [about a dozen releases](https://github.com/anthropics/claude-code/blob/main/CHANGELOG.md) in the two weeks after mods launched. Expect to fix your mods after updates.

## Should you use them

Yes. It’s only been a few days and all I have is my own experience, but mods have already made my workflow better and made me trust my agents more.

If you keep repeating the same rules to your agent, start with those. My rule of thumb: if a rule in CLAUDE.md exists because something went wrong on a specific day, make it a mod. Keep CLAUDE.md for judgment calls like tone and taste, where there’s nothing for code to check. And keep it small. I wrote up [what I kept always-on in CLAUDE.md after splitting it](https://www.joekarlsson.com/blog/claude-md-open-knowledge-format/) Oct 30, 2026**Can Open Knowledge Format make Claude Code more accurate and cheaper? I tested it** .

Keep any hooks that block things, but check that they actually block. And if you have hooks that feed Claude extra context, test them: have one tell Claude a secret word, then ask for it.

## Works on My Machine

Get the next post by email. My newsletter covers homelab builds, AI tools and open source, and it goes out when I have something to say.

It’s free, and you can unsubscribe anytime. You’ll get an email to confirm first.

![Joe Karlsson at his desk, chin on hand, with a microphone in front of him](https://www.joekarlsson.com/images/about/collage/studio.webp)

Joe  
Karlsson

Technology is better when it's weird and it's yours. Everything I write about runs in my house.

[More about me →](https://www.joekarlsson.com/about/)

[Bluesky](https://bsky.app/profile/joekarlsson.com) | [LinkedIn](https://www.linkedin.com/in/joekarlsson/) | [GitHub](https://github.com/JoeKarlsson)
