JuliusBrussee/

caveman

GitHub Website

Caveman is a JavaScript skill/plugin designed to reduce output tokens in AI responses by 65% while maintaining technical accuracy. It supports over 30 agents and offers various installation methods and usage commands.

88kstars

5.1kforks

397issues

MITlicense

2026since

Star historydaily snapshots by VibeCrowd

Collecting history — the radar snapshots this repo daily. The trend line appears after 3 days of data (1 so far).

Alternatives & relatedmatched by topic overlap

headroomlabs-ai/

headroom

Compress tool outputs, logs, files, and RAG chunks before they reach the LLM. 60-95% fewer tokens, same answers. Library, proxy, MCP server.

59k4.3kPython

asgeirtj/

system_prompts_leaks

Extracted system prompts from Anthropic - Claude Fable 5, Opus 4.8, Claude Code, Claude Design. OpenAI - ChatGPT GPT-5.6, Codex GPT-5.6, GPT-5.5. Google - Gemini 3.5 Flash, 3.1 Pro, Antigravity. xAI - Grok, Cursor, Copilot, VS Code, Perplexity, and more. Updated regularly.

56k9.3kJavaScript

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.

229k35kJavaScript

DietrichGebert/

ponytail

Makes your AI agent think like the laziest senior dev in the room. The best code is the code you never wrote.

81k4.4kJavaScript

NousResearch/

hermes-agent

The agent that grows with you

213k39kPython

thedotmack/

claude-mem

Persistent Context Across Sessions for Every Agent – Captures everything your agent does during sessions, compresses it with AI, and injects relevant context back into future sessions. Works with Claude Code, OpenClaw, Codex, Gemini, Hermes, Copilot, OpenCode + More

87k7.5kJavaScript

Comparecaveman vs headroom caveman vs system_prompts_leaks caveman vs ECC

Reviewgenerated from repository data · Jul 12, 2026

What it is

Caveman is a skill/plugin for Claude Code, Codex, Gemini, Cursor, Windsurf, Cline, Copilot, and over 30 other agents. It reduces the number of output tokens by 65% while preserving the technical accuracy of responses.

How it works

Install drops a skill file into your agent.
Skill tells agent: drop filler, keep substance, use fragments — but never touch code, commands, or errors.
On Claude Code, a hook writes a tiny flag file each session, so the agent talks caveman from message one without /caveman.
/caveman-stats reads your session log, counts tokens saved, writes the number to your statusline.
/caveman-compress rewrites memory files (like CLAUDE.md) so every future session starts with a smaller context. Save tokens forever, not just once.

Getting started

To install Caveman, use the following command:

# macOS · Linux · WSL · Git Bash
curl -fsSL https://raw.githubusercontent.com/JuliusBrussee/caveman/main/install.sh | bash

For Windows users, the command is:

# Windows · PowerShell 5.1+
irm https://raw.githubusercontent.com/JuliusBrussee/caveman/main/install.ps1 | iex

After installation, activate it by typing /caveman or saying "talk like caveman". To deactivate, say "normal mode".

Recent releases

v1.9.1 (2026-07-03): A maintenance + honesty release with 50 changes, mostly related to hook, installer, and integration hardening.
v1.9.0 (2026-06-12): Security improvements including integrity-checked installs.
v1.8.2 (2026-05-12): Installer bug fixes.
v1.8.1 (2026-05-10): Hotfix for the broken install command in v1.8.0.
v1.8.0 (2026-05-10): Initial release of the curl|bash one-liner command.

Traction

Caveman has 88158 stars and 5064 forks on GitHub.

Caveats

License: MIT
Open issues: 397
Age: Created on 2026-04-04, last pushed on 2026-07-03.

SharePost on X LinkedIn