logo
Kurashizu Blogwhere ideas flow
  • Home
  • News
  • Blog
  • About
  • Home
  • News
  • Blog
  • About

© 2026 Kurashizu. All rights reserved.

Admin·New Post·Service Status
Powered by Cloudflare·01.11.26
← Back to news

How to stop Claude from saying load-bearing

Original article
jola.dev|197 points|301 comments|by shintoist|Jul 14, 2026

Eliminating "Load-Bearing" from Claude's Vocabulary

Are you losing your mind completely exhausted from reading Claude's obsession with "load-bearing seams" and its tendency to offer "honest takes" on every single topic?

If you can't stop the AI from being repetitive, the only logical solution is to make the output so absurd that you can't help but laugh.

The Strategy

To achieve this, we will implement a replacement script that intercepts the output. We can quantify the "Annoyance Factor" (A\mathcal{A}A) of Claude's prose using the following formula:

A=∑(load-bearing+honest take)×frustration2\mathcal{A} = \sum (\text{load-bearing} + \text{honest take}) \times \text{frustration}^2A=∑(load-bearing+honest take)×frustration2

To bring A\mathcal{A}A down to zero, follow these steps:

1. Create the Swap Script

First, you need a Python script to handle the text substitutions.

#!/usr/bin/env python3
import json, re, sys

# Mapping the annoying phrases to ridiculous alternatives
replacements = {
    "seam": "whatchamacallit",
    "you're absolutely right": "I'm a complete clown",
    "honest take": "spicy doodad",
    "load-bearing": "cooked"
}

# Constructing the hook response
data = json.dumps({
    "hookSpecificOutput": {
        "hookEventName": "MessageDisplay",
        "displayContent": text, # Assuming 'text' is captured from input
    }
})

2. Installation & Permissions

You must place this script in a specific directory and ensure the system can execute it.

  • Path: ~/.claude/hooks/wordswap.sh
  • Command: chmod +x ~/.claude/hooks/wordswap.sh

3. Configuration

Now, link the script to your settings. Edit your ~/.claude/settings.json file to include the following hooks block:

{
  "hooks": {
    "MessageDisplay": [
      {
        "hooks": [
          {
            "type": "command",
            "command": "$HOME/.claude/hooks/wordswap.sh"
          }
        ]
      }
    ]
  }
}

AI Confusion

Implementation Workflow

The following diagram illustrates how the message is intercepted:

Summary of Changes

Here is a quick reference for what is being swapped:

Original PhraseNew Ridiculous Phrase
seamwhatchamacallit
you're absolutely rightI'm a complete clown
honest takespicy doodad
load-bearingcooked

Final Checklist

  • Script written and saved to ~/.claude/hooks/
  • Permissions set to executable
  • settings.json updated with the MessageDisplay hook
  • New session started (hooks load at startup)

Feel free to experiment with your own replacements to make the experience even more productive—or more chaotic.


Support the Author If you enjoyed this, consider supporting my work:

  • 🙏 Sponsor me on GitHub
  • ☕ Buy me a coffee on Ko-fi

Other Reading

  • Let libraries be libraries (July 07, 2026): A critique of Elixir applications acting as libraries.
  • CI workflows on Tangled for Elixir (July 04, 2026): Guide for setting up PostgreSQL and specific Erlang/Elixir versions.
  • atproto tangled (June 29, 2026): A project for syncing blog feeds to standard.site and atproto.