Companies Use 'Caveman' Plugin to Force AI Tools to Give Terse Responses and Cut Token Costs
By
Joseph Cox
Summary
Companies are using a 'caveman' plugin for AI coding tools like Claude Code, Codex, and Gemini to force them to give extremely short, terse responses (e.g., "Hulk smash" instead of verbose explanations) in order to reduce token usage and curb skyrocketing AI costs. The plugin has even received contributions from a senior OpenAI employee. This reflects growing corporate anxiety over unpredictable and massive AI expenditure.
Source
Key quotes
· 5 pulledCompanies are deliberately making their AI tools speak like cavemen in an attempt to stop burning through AI tokens and curb their massive expenditure on AI
The tool turns the usually verbose outpost of LLMs like Claude Code, Codex, or Gemini into a much more to the point answer
Think less 'you're right to push back, I was wrong,' and more 'Hulk smash.'
Use of the caveman plugin is in direct response to the skyrocketing and unpredictable cost of AI
A senior OpenAI employee has contributed code to the project, simply called 'caveman.'
You might also wanna read
Caveman Plugin Reduces Claude Output Tokens by 75% While Maintaining Accuracy
Caveman is a Claude Code skill plugin that reduces Claude's output tokens by approximately 75% while maintaining technical accuracy. It feat
Benchmarking Caveman compression plugin: "Be brief" achieves same results
A developer benchmarks the Caveman Claude Code compression plugin against simply using the prompt "be brief." The findings show that Caveman
Research Study: AI Coding Assistants' Tool Recommendations Analysis
A research study analyzing AI coding assistants' tool recommendations by testing Claude Code on real repositories 2,430 times. The study exa
The Rise of AI Coding Agents and the 'Token Anxiety' They Create in Software Development
The article discusses the author's growing anxiety about the rise of AI coding agents and their impact on software development culture. The
AI Coding Tools Create Review Burden for Open Source Maintainers
The article discusses the challenges that AI coding tools like GitHub Copilot, Cursor, and Claude create for open source maintainers. While
samsaffron.com·8mo agoBuilding a Production App with 150,000 Lines of AI-Generated Elixir Code: Lessons Learned
The article details the experience of developing BoothIQ, a lead retrieval app for trade shows, using 150,000 lines of Elixir code written e

Comments
Sign in to join the conversation.
No comments yet. Be the first.