Back to directory

Caveman Prompt File for GitHub Copilot

A lightweight prompt file for GitHub Copilot that makes the agent respond in terse caveman-style speech, cutting a large share of output tokens while keeping technical accuracy, without installing the original package.

Caveman Prompt File for GitHub Copilot

This project packages a slash-command prompt file for VS Code and a matching always-on instructions file, based on the original Caveman project's observation that terse, caveman-style phrasing cuts around 75 percent of output tokens, and, with an added mode for classical Chinese phrasing, about 46 percent of input tokens as well, without losing technical substance.

VS Code reads these prompt files directly, while JetBrains and most other IDEs instead need the prompt text copied into Copilot's custom instructions. Make targets install the prompt files globally, so the slash command works in every workspace, or into one specific repository, in either an on-demand or an always-on mode, copying the checked-in files rather than downloading them again. It is licensed MIT.

More in Agent skills

See all 581
  • Provenance gate

    github.com

    A deterministic authorization gateway that sits between an AI agent and its tools, tracing where each parameter of a side-effect call came from, and blocking the call or asking a human when the source does not check out.

  • A markdown framework of system rules that restates Isaac Asimov's Four Laws of Robotics as a set of prioritized loss calculations for large-language-model agents to follow, published as a discovery file for agents and crawlers to read.

  • Vitruvius

    github.com

    An engineering research agent that runs a discover, read, synthesize, verify and review loop over technical questions, reading sources directly rather than inferring from titles, and publishing its own scored verifier benchmark, weaknesses included.