# Caveman

> Viral skill and proxy that forces coding agents to spend fewer tokens

**Category:** AI Agents  
**Pricing:** open-source  
**URL:** https://www.caveman.ai  
**GitHub:** https://github.com/JuliusBrussee/caveman  
**Added:** 2026-09-20

---

## The Hook

Caveman is the skill-plus-proxy stack that makes coding agents talk tersely so token bills shrink without swapping models.

## What's Good

MIT skill packaging (JuliusBrussee/caveman) with install paths across many agent hosts, plus middleware/CLI wrappers that rewrite verbose agent chatter into short caveman-style output. Independent writeups (including JetBrains and an Adobe Research CAVEWOMAN paper cited in-repo) treat the compression as real cost leverage, not only a meme.

## The Catch

Engine-linked runtime directories (engine/, proxy/, rewriter/, and related) are Business Source License 1.1, not MIT — see LICENSING.md / LICENSE.BSL before embedding or offering a hosted Caveman runtime. Vs LiteLLM: Caveman is output/style compression for agent hosts, not a multi-provider model gateway. Terse output can hide nuance; keep quality checks on hard tasks.

## Verdict

MIT skill surface with BSL engine bits. Use to cut coding-agent token spend; read the BSL boundary before productizing the runtime.