Vibe Coding Discovercaveman
View on GitHub🪨 why use many token when few token do trick. Viral skill + proxy for coding agents that cuts 65% of tokens by talking like a caveman.
A cross-agent skill, local proxy, and middleware that reduce token use by compressing agent prose and tool results while preserving originals for retrieval. Supports 30+ coding agents and integrations including LangChain and Vercel AI SDK.
Use Cases
Reduce coding agent response tokensCompress tool output before it reaches an LLMLower token usage in custom AI applicationsReview code with concise findingsSummarize command output and web pages
Built With
- Language
- Go
- Frameworks
- LangChain · Vercel AI SDK
Tags
agent skill · token reduction · prompt engineering · coding agents · output compression · input compression · proxy · middleware · Claude Code · Codex