Compress natural language memory files into caveman format to reduce tokens.
npx skills add https://github.com/DDTully/dotfiles --skill compress-ddtully
Please help me install this Agent Skill. Skill: compress Source: https://github.com/DDTully/dotfiles/tree/main/skills/.agent_skills/compress Command: npx skills add https://github.com/DDTully/dotfiles --skill compress-ddtully
Compress memory files (CLAUDE.md, todos, preferences) into caveman format to reduce per-session token usage while preserving content and structure.
Run the caveman compress process on a target memory file to overwrite the original with a compressed version and generate a human-readable backup.
High-intent search queries and answers about installing and using this skill.
To reduce Claude memory tokens, compress natural language files into a condensed caveman format. This minimizes per-session token usage while preserving essential code blocks, URLs, and file paths.
Your original markdown files are preserved by creating a human-readable backup named <filename>.original.md. The active memory file is then overwritten with the newly compressed version.
Yes, the markdown memory compression workflow requires the anthropic and tiktoken Python dependencies. These drive the CLI workflow that detects language content and validates the compressed output.
The validation process checks compressed Claude memory files and performs limited retries with targeted fixes if issues are detected. This ensures the final caveman format output maintains structural integrity.
Yes, you can apply this token compression technique to project-level todos and preferences. The CLI-driven workflow detects language content across various memory files to systematically reduce token load.
jchildreeCompress natural language text from markdown files using Claude API and Python.
shirulotCompress natural language in Markdown files while preserving code blocks and file paths.
nnexaiManage project phase CRUD operations in ROADMAP.md files.
tassiovaleCompress natural language markdown files into caveman format.
