10 alternatives to caveman

juliusbrussee/caveman — 🪨 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. (114,191 stars, Apache-2.0, ai, dev-tools, auth).

These are the closest projects in the JustClaw-It index that we serve licence-cleared code from, ranked by relevance and by our code-quality score (1–10, the mean over each repo's graded components: completeness, code quality, deployability, reusability, strategic value). JustClaw-It also serves 477 licence-cleared components from juliusbrussee/caveman itself.

The alternatives in detail

• supersynergy/agent-token-saver
  Cuts what your coding assistant spends per task, and shows you the numbers. Works with Codex, Claude Code, Hermes, GG Coder and any command-line agent.
  score 5.6/10 · MIT permissive · 8 stars · 43 components · search them
• headroomlabs-ai/headroom
  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.
  score 7.5/10 · Apache-2.0 permissive · 77,596 stars · 816 components · search them · alternatives
• pacifio/atlas
  Source control for agents. Use multiple coding agents, track their changes and query them in one place
  score 6.6/10 · Apache-2.0 permissive · 9,621 stars · 1,165 components · search them
• conorbronsdon/avoid-ai-writing
  Skill that audits and rewrites content to remove AI writing patterns. Use it with your favorite agents including Claude Code, OpenClaw, Codex, and Hermes.
  score 5.7/10 · MIT permissive · 5,166 stars · 50 components · search them
• greyhaven-ai/autocontext
  a recursive self-improving harness designed to help your agents (and future iterations of those agents) succeed on any task
  score 6.0/10 · Apache-2.0 permissive · 1,310 stars · 1,910 components · search them
• wenyuchiou/awesome-agentic-ai-zh
  A trilingual (繁中 / English / 简中) learning roadmap for agentic AI: from LLM basics to multi-agent systems, with 240+ curated resources and hands-on examples. 中文 AI agent 學習地圖。
  score 6.2/10 · MIT permissive · 7,901 stars · 36 components · search them
• leonxlnx/unlazy
  Anti-laziness skill for AI agents. Core: the Depth Tree method, which splits a task N layers deep and gives every leaf the full time budget of the whole task, so effort multiplies with depth. Grounded in 2025-2026 resear
  score 6.3/10 · MIT permissive · 4,077 stars · 9 components · search them
• can1357/oh-my-pi
  ⌥ Coding agent with the IDE wired in. Built by Stencil Labs.
  score 6.2/10 · MIT permissive · 36,241 stars · 3,000 components · search them · alternatives
• affaan-m/ecc
  The agent harness performance optimization system. Skills, instincts, memory, security, and research-first development for Claude Code, Codex, Opencode, Cursor and beyond.
  score 6.8/10 · MIT permissive · 286,740 stars · 667 components · search them · alternatives
• dietrichgebert/ponytail
  Makes your AI agent think like the laziest senior dev in the room. The best code is the code you never wrote.
  score 6.7/10 · MIT permissive · 175,543 stars · 37 components · search them · alternatives

Licence data: JustClaw-It's per-repo licence policy. Is caveman still maintained? Check it. Licence details: licence checker. Ranking data from 2026-10-29T06:21:36Z.