gitwtfhub

wtf is cutweight?

brenbuilds1/cutweight — explained in plain English

Analysis updated 2026-05-18

2Audience · developerSetup · easy

TL;DR

Five small skills for AI coding assistants that measure and trim context bloat from installed skills, memory files, replies, and long sessions.

Mindmap

mindmap
  root((cutweight))
    What it does
      Measures context bloat
      Trims replies
      Manages sessions
    Tech stack
      SKILL.md format
      Claude Code plugin
      npx CLI
    Use cases
      Weigh in bloat
      Skill audit
      Reply trimming
    Audience
      AI assistant users
      Context conscious developers

Code map

Detail Auto

An interactive map of this repo's files and how they connect — its source is parsed live in your browser. Click Visualize to build it.

filefunction / class

Why would anyone build with this?

REASON 1

Measure how much always-on context your AI assistant setup is carrying with Weigh In.

REASON 2

Get a keep, trim, or cut recommendation table for your installed skills and memory files.

REASON 3

Trim filler from an AI assistant's replies while keeping exact commands and error messages.

REASON 4

Manage long coding sessions with a clean handoff note before resetting context.

What's in the stack?

SKILL.mdClaude CodeCodexCursorNode.js

How it stacks up

brenbuilds1/cutweight0-bingwu-0/live-interpreter010zx00x1/faresnipe
Stars222
LanguagePythonPython
Setup difficultyeasymoderateeasy
Complexity2/52/5
Audiencedevelopergeneralgeneral

Figures from each repo's GitHub metadata at analysis time.

How do you spin it up?

Difficulty · easy Time to first run · 5min

The npx CLI install needs Node.js 18 or newer, otherwise use the plugin or manual copy paths.

Wtf does this do

Cutweight is a set of five small skills that plug into AI coding assistants like Claude Code, Codex, or Cursor, aimed at trimming the extra content those assistants read on every single message. The idea is that every skill installed, every memory file kept around, and every tool schema loaded adds weight to the context window, and that weight makes the assistant slower to respond well, not just more expensive to run. The first skill, Weigh In, measures how much always-on content your setup is carrying and gives you a weight class rating along with a shareable summary card, so you can track whether things get lighter or heavier over time. Skill Diet goes further and reads through everything installed, including skills, memory files, and tool schemas, and tells you what is always costing tokens versus what only fires occasionally versus what never runs at all, without deleting anything itself. Word Diet changes how the assistant replies, cutting filler out of its answers while keeping exact details like commands, code, and error messages intact. Think Light looks at your tasks and sorts them into light and heavy piles, only reducing how much the assistant reasons through the light ones after checking that quality does not suffer. Session Cut focuses on managing long conversations, encouraging one task per session and giving a short handoff note before starting fresh, so a bloated conversation history does not start working against you. Installing it is done through a command line tool called skills, or as a Claude Code plugin, or by manually copying the skill folders into your assistant's skills directory. The author notes plainly that these five skills themselves add a small always-on cost of roughly 700 tokens to your setup, an honest disclosure rather than a hidden cost.

Yoink these prompts

Prompt 1
npx skills add brenbuilds1/cutweight and run the weigh-in skill on my current setup
Prompt 2
Run skill-diet and show me which of my installed skills never actually fire
Prompt 3
Apply word-diet so your replies are shorter but keep all commands and error text exact
Prompt 4
Use think-light to sort my current task list into light and heavy piles
Prompt 5
Walk me through the session-cut handoff note before I reset this conversation

Frequently asked questions

wtf is cutweight?

Five small skills for AI coding assistants that measure and trim context bloat from installed skills, memory files, replies, and long sessions.

How hard is cutweight to set up?

Setup difficulty is rated easy, with roughly 5min to a first successful run.

Who is cutweight for?

Mainly developer.

View the repo → Decode another repo

This repo across BitVibe Labs

Don't trust strangers blindly. Verify against the repo.