LLM token reduction via prompt engineering
Claude Code skill that reduces token usage by rewording prompts in a terse 'caveman' style, cutting input tokens by ~33-65%.
Solves the same problem with a different architecture or at a different layer. Expect to rewrite the integration.
Choose it when you want algorithmic compression of tool outputs, logs, and RAG chunks rather than a prompt-style change; works with any LLM integration, not just Claude Code skills.
Compress tool outputs, logs, files, and RAG chunks before they reach the LLM. 20% fewer tokens for coding agents, 60-95% fewer tokens for JSON, same answers. Library, proxy, MCP server.
These projects were analysed and named caveman among their alternatives. The relationship is not symmetric — how caveman rates them is a separate judgement, made when caveman is analysed in its own right.
calls caveman “Choose it when you want a Claude Code skill that changes language style to cut tokens, no library/proxy.”
Compress tool outputs, logs, files, and RAG chunks before they reach the LLM. 20% fewer tokens for coding agents, 60-95% fewer tokens for JSON, same answers. Library, proxy, MCP server.