caveman-compress

Compress Markdown memory files into caveman format with backups.

99|6|Updated Jan 11, 2026
One-click install
npx skills add https://github.com/vvedantb/eva --skill caveman-compress-vvedantb
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: caveman-compress
Source: https://github.com/vvedantb/eva/tree/main/.agents/skills/caveman-compress
Command: npx skills add https://github.com/vvedantb/eva --skill caveman-compress-vvedantb

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires anthropic, tiktoken, and includes scripts (resource) components.

What problem does it solve?

Compress natural language memory files (CLAUDE.md, todos, preferences) into caveman format to reduce input tokens. Compressed version overwrites the original file. Human-readable backup saved as FILE.original.md.

Core Features & Use Cases

  • Compress memory files to reduce token load on session start
  • Maintain a backup as FILE.original.md and provide an option to re-compress after edits
  • Validate and retry on errors to ensure correctness

Quick Start

Compress memory file by running caveman-compress on the target file to replace it with a caveman-compressed version and save the original as a .original.md backup

Frequently Asked Questions about caveman-compress

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I reduce token usage when loading large markdown memory files in Claude?

Compressing markdown memory files into caveman format reduces the token load on session start. The tool overwrites the original file with the compressed version and saves a human-readable backup as FILE.original.md for safety.

What is the caveman format for compressing markdown memory docs?

Caveman format is a compressed representation of natural language markdown content used to minimize input tokens. It transforms detailed memory notes into a shorter, stripped-down syntax while preserving essential context for Claude.

How do I compress CLAUDE.md files without losing the original content?

You can compress CLAUDE.md files safely by running a compression tool that creates a backup. The process saves your original content as a FILE.original.md backup before overwriting the active memory file with the compressed version.

Does markdown memory compression work with tiktoken and anthropic dependencies?

Yes, markdown memory compression relies on anthropic and tiktoken dependencies to process and validate token counts. It applies deterministic CLI orchestration to handle per-file compression and optional Claude API calls.

What happens if errors occur during markdown memory file compression?

If errors occur during compression, the process implements validation and retry loops to ensure correctness. This safe backup workflow ensures your original markdown content is preserved before applying the compressed file.