{
  "slug": "caveman",
  "name": "caveman",
  "url": "https://github.com/JuliusBrussee/caveman",
  "cat": "cost",
  "group": "claude",
  "tagline": "Token-compression skill + tooling",
  "price": "Free",
  "tier": "free",
  "pick": false,
  "what": "A skill that makes the agent respond in terse fragment style to cut output tokens, at four compression levels. Companion tools attack the input side: <code>caveman-compress</code> rewrites CLAUDE.md and memory files (~46% fewer input tokens), <code>caveman-shrink</code> wraps any MCP server with compressed tool descriptions. ~69k stars.",
  "why": "The <code>caveman-compress</code> and <code>caveman-shrink</code> utilities are the parts worth having — they save tokens without making the agent's prose unreadable.",
  "warn": "Headline claim is 75% output reduction; the repo's own 10-task benchmark reports 65% average with a 22–87% range. And terse fragment output is genuinely harder to review.",
  "install": "curl -fsSL .../install.sh | bash",
  "category": {
    "key": "cost",
    "name": "Cost & efficiency",
    "desc": "Same answers, fewer tokens. Measurement first, compression second.",
    "group": "claude",
    "groupName": "Claude Code toolkit"
  },
  "freshness": null,
  "health": {
    "ok": true,
    "status_code": 200,
    "final_url": null,
    "checked_at": "2026-08-18T20:38:44.869+00:00"
  },
  "corrected_at": null,
  "upstream_changed_at": null
}