In the coding agent you run
Claude Code, Codex, Cursor, Gemini and 30+ more. One command each, on your own keys.
- says less
npx skills add JuliusBrussee/caveman- reads less
npx @caveman-ai/cli
Caveman makes every agent do the same job for less. Free on your machine, and measured for your team in Cloud.
For your team. See what your agents cost and cut it, hosted or in your VPC.
freeApache-2.0runs on your machineno account108k stars on GitHub
Shorter replies with the same intent, in every agent that loads skills.
Logs, JSON, code and diffs shrink before the model reads them. The model can ask for the original.
A page reaches the agent as the content it needs, not a full accessibility snapshot.
Numbers measured on your machine are labelled inferred. Only Cloud reports verified savings.
Claude Code, Codex, Cursor, Gemini and 30+ more. One command each, on your own keys.
npx skills add JuliusBrussee/cavemannpx @caveman-ai/cliTypeScript and Python. You keep your framework, your provider and your history.
npm install @caveman-ai/[email protected] @caveman-ai/[email protected]pip install "caveman-sdk==1.1.0" "caveman-middleware[langchain]==0.1.0a1"caveman learn ranks the worst waste in your sessions and fixes it, one diff at a time.free to starthosted or your VPCnever priced per seat

Sample workspace, fictional data. Open the demo
Bring in Cloud when your team wants every saving measured.
No account Your own keys Open source
Earlier tools, still on GitHub without active development: Caveman Code, Cavemem, Cavekit.