By yibie
Cut token usage by ~75% in Claude Code interactions, commit messages, and code reviews using ultra-compressed caveman-style speech that preserves technical accuracy.
Ultra-compressed commit message generator. Cuts noise from commit messages while preserving intent and reasoning. Conventional Commits format. Subject ≤50 chars, body only when "why" isn't obvious. Use when user says "write a commit", "commit message", "generate commit", "/commit", or invokes /caveman-commit. Auto-triggers when staging changes.
Ultra-compressed code review comments. Cuts noise from PR feedback while preserving the actionable signal. Each comment is one line: location, problem, fix. Use when user says "review this PR", "code review", "review the diff", "/review", or invokes /caveman-review. Auto-triggers when reviewing pull requests.
Ultra-compressed communication mode. Cuts token usage ~75% by speaking like caveman while keeping full technical accuracy. Supports intensity levels: lite, full (default), ultra, wenyan-lite, wenyan-full, wenyan-ultra. Use when user says "caveman mode", "talk like caveman", "use caveman", "less tokens", "be brief", or invokes /caveman. Also auto-triggers when token efficiency is requested.
../../caveman-compress/SKILL.md
Codex-first port of caveman. Install as a Codex Desktop local plugin and trigger with $caveman.
This repository adapts the original caveman project for OpenAI Codex Desktop local-plugin installation.
For the full project overview, examples, features, and benchmarks, see the upstream README:
Run inside the project you want to enable Caveman for:
curl -fsSL https://raw.githubusercontent.com/yibie/caveman-codex/main/install.sh | bash
Or pass the target project explicitly:
curl -fsSL https://raw.githubusercontent.com/yibie/caveman-codex/main/install.sh | bash -s -- --project "/path/to/your/project"
The installer:
<project>/.codex-plugins/caveman<project>/.agents/plugins/marketplace.json to reference ./.codex-plugins/cavemanThen restart Codex Desktop and install caveman from the Plugins UI.
.commandIf you cloned this repo locally:
/path/to/caveman-codex/install-codex-plugin.command "/path/to/your/project"
You can also double-click install-codex-plugin.command in Finder and choose the target folder.
$caveman$caveman lite, $caveman full, $caveman ultrastop caveman or normal modemarketplace.json exists.Own this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimOwn this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimBased on adoption, maintenance, documentation, and repository signals. Not a security audit or endorsement.
npx claudepluginhub yibie/caveman-codex --plugin cavemanUltra-compressed communication mode. Cuts 65% of output tokens (measured) while keeping full technical accuracy by speaking like a caveman.
超圧縮コミュニケーションモード。原始人のように話してトークン使用量を約75%削減しつつ、技術的正確性は完全に維持。日本語に最適化。コミット生成・PRレビュー・メモリ圧縮・stats可視化サブスキル同梱。SessionStart/UserPromptSubmit フックでモード追跡・毎ターン補強・ドリフト防止・/genshijin-stats でセッション削減量+USD推定表示。MCP middleware (genshijin-shrink) でMCPツール記述も圧縮。スラッシュコマンド /genshijin /genshijin-commit /genshijin-review /genshijin-stats 付属。Cursor/Windsurf/Cline/Copilot 等マルチエージェント対応。3 cavecrew相当 subagent (investigator/builder/reviewer) で長セッションコンテキスト持続。
Verdict-first communication mode. Cuts output tokens while keeping full technical accuracy.
Token optimization for Claude Code. Automatic tool output compression (40-60% reduction), token meter in statusline, auto-compact at 70% context, structured output formats, self-building project wiki, URL ingestion, Karpathy compile pass, and semantic lint.
Headroom startup hooks for Claude Code and GitHub Copilot CLI.
Ультра-сжатый режим общения на русском и казахском (Қысқа сөйлесу режимі қазақша/орысша). Сокращает ~65-80% токенов при сохранении полной технической точности. Пещерный режим для Claude Code.