docs(readme): say what Caveman Code measured
Drop the 76% working-day cell (3 sessions, run once; the matched GPT-6.1 Sol run was a loss). Terminal-Bench 4 against Claude Code is level within noise (5.19vs6.04 per solved task, 95% interval 0.70x to 1.02x), not 14% cheaper. The tagline becomes the measured line, 21% less than Codex per solved task on Terminal-Bench 4, and the figures say they come from the benchmark build, not the shipped app.
Co-Authored-By: Claude Opus 5.5 noreply@anthropic.com
版权所有:中国计算机学会技术支持:开源发展技术委员会
京ICP备13000930号-9
京公网安备 11010802047560号
New: Caveman Code. Free.
21% less than Codex per solved task on Terminal-Bench 4. Same model on both sides.
21% cheaper
than CodexTerminal-Bench 4, per solved task, GPT-6.1 Sol: $1.21 vs $1.52, 90 of 192 runs solved both
Level
with Claude Code within noiseTerminal-Bench 4, Opus 5.5: 97 of 180 runs solved vs 83 of 177; $5.19 vs $6.04 per solved task, 95% interval 0.70x to 1.02x
Measured build, benchmark settings on, not the shipped app: the app ships the same harness with three benchmark-only settings turned off. Same model on both sides: GPT-6.1 Sol against Codex, Opus 5.5 against Claude Code. Our own runs, November 2026, at list prices. How we measured
Every coding agent in one window. Free on the subscriptions and keys you already have.
Download Caveman Code for Mac · Read the launch post
why many token when few do trick
Caveman make your AI agent say less and read less. Code stay exact. Brain still big.
33.2% fewer
input tokens, whole sessions6 synthetic large-tool-output tasks, 18 pairs, Aug 2026 engine, 18 of 18 answers right
0.46×
the cost of Opus at xhigh, with Auto85 coding tasks, interim, score gap ±7 points, unresolved
397 → 76
tokens for one test runBlocks vs
| tail -201.4 to 2.4× cheaper
with caveman-style outputAdobe Research, eight models
Cited by Adobe Research · A/B tested by JetBrains · Remade for Elasticsearch on Elasticsearch Labs
#1 on Hacker News · #1 on GitHub Trending · “No way this actually works.” ThePrimeagen
Install · Six rocks · Proof · Say less · Read less · Auto · Blocks · Commands · Caveman Cloud
Same fix. 63 token become 20. Brain still big.
Pick your club:
/cavemanuseMemo./ultracaveuseMemo./megacaveuseMemo。Token counts: tiktoken o200k.
Install
One command. Mac, Linux, Windows. Caveman find your agents and your logins, show what it change. Press Enter. Done. Needs Node.js 22.13+.
After that, plain
claudestays caveman’d. So do Codex, Gemini CLI, OpenCode, and 30+ more. It runs on your machine, with your keys and your Claude Pro/Max login. One rock. That it.Had caveman before 4.0? Say
@latest, or the old copy on your machine wins. Already set up?caveman setupopens the screen again. Only want shorter answers? Get just the skill.Six rocks, one install
New in 4.0: caveman is six parts. All on after install. Each has its own switch.
caveman statusshows what is on.caveman off browseturns one off. Only Auto needs an account, and it is free. 4.0.0 release notesProof
Outside labs first. Ours come with each rock below. Nothing rounded up. Red rows stay red.
Two findings shaped caveman. Adobe: cavemanning your prompt makes answers worse, so caveman never touches your prompt. JetBrains: agent sessions are mostly code and tool calls, which the voice leaves alone. So caveman grew rocks that shrink what the agent reads and what it runs on.
Say less: the caveman voice
A voice, not broken grammar. Every reply follows the same structure.
[thing] [action] [reason]. [next step].No greeting, no “let me”, no recapExample run, n=10, not a claim. Ten dev questions on claude-opus-5-5:
Answer concisely./caveman/ultracave“Be concise” is the real baseline; new models already know it. On top of it,
/cavemancuts 3% more at the median and/ultracave35% more. Measured on the 3.1.0 rules, not yet re-run. The rules cost about 1,160 tokens of input, and if you pay per request (GitHub Copilot), shorter answers save nothing. Harness · Every case where caveman losesOnly want this rock? The skill alone, no proxy:
Type
/caveman($cavemanin Codex) if it doesn’t start on its own. Saystop cavemanto go back.Other ways in: Claude app and phone, Claude Code plugin, Gemini, every agent at once, Windows, uninstall
Claude on the web, the phone app, or Cowork? No terminal. Add caveman to your Claude account once, in Customize > Plugins: steps. Not tested end to end yet.
Already inside Claude Code? Same rock, slash form. Type these where you talk to Claude:
Windows, PowerShell 5.1+:
On npm 12 or newer, new npm block git install. One-liners above handle it themselves from v3.2.0 on. Old one-liner pinned to v3.1.0 or older still fail there: use one above, or this:
npx --allow-git=root -y github:JuliusBrussee/caveman#v4.1.0Changed your mind:
npx -y github:JuliusBrussee/caveman -- --uninstall(on npm 12+, add--allow-git=roottoo)Only grabbed the skill?
npx skills remove JuliusBrussee/caveman -gtake it back out.Install broke? Open your agent in this repo and say “Read INSTALL.md, install caveman for me.” Agent fix own brain.
Read less: 33.2% fewer input tokens
The voice shrinks what the agent says. This shrinks what it reads. Same answers: 18 of 18 right. 6 synthetic large-tool-output tasks, 18 pairs, Aug 2026 engine.
The file is each test file through today’s compressor alone. The session is the whole agent run, which also carries the system prompt and tools, so it moves less. Honest part: the session run is from August 2026 on an older engine, HTML has no compressor yet so it lost, and each test file hides one record in 61 to 72 KB of noise. Yours will vary. Headroom on the same suite: 15 of 18 right, 6.7% fewer tokens (interval crosses zero; harness not public). Method
Every original stays on your disk. The agent can pull it back any time.
Auto: 0.46x the cost of Opus at xhigh (interim)
New model in your agent’s picker. Pick Auto and caveman chooses the model and the effort for each ask. Hard ask, big brain. Easy ask, save rocks.
Auto costs 0.46× what Opus 5.5 at xhigh costs. That is 54% saved, in under half the time. 85 real coding tasks from SWE-Interact and SWE-Together, one try each. Honest part: the run is half done, the 0.5-point score gap is too small to tell apart at this size (±7), and most of the saving is Auto lowering effort on easy asks. Already on Opus at medium? Expect less; not measured yet. Full table and limits
Works in Claude Code, OpenCode, and Codex on a ChatGPT login, with API keys, Claude Pro/Max, or ChatGPT. Any other model you pick runs as picked. Not signed in: no Auto, nothing sent. What Auto sends
Blocks: code mode, script written once
skills.sh for code mode. Code mode is the agent writing a script instead of making tool call after tool call (Cloudflare, Anthropic). Problem: agent write same script every session, throw it away, write it again. Blocks keep it in your repo, check it every pull request, hand it back.
go test -v ./..., 326 tests| tail -20test-summaryjson-peekgrep -ngrep-defsTokens the agent reads back. In one week, agents on one machine wrote the same script 394 times. Purdue measured the habit at up to 9.57% of task cost.
grep-defsloses to a bare grep; row stays. MethodComes with caveman. Eight blocks ready:
json-peek·jsonl-stats·test-summary·first-error·wait-for·grep-defs·http-json·replace-in-file. When the agent starts typing a script a block already covers, it gets one line: use the block. Never denies or rewrites a command.Pre-release (v0.1.0-rc.5). Python only, no sandbox yet. Own repo, own stars: caveman-ai/blocks
Smaller rocks
caveman learncaveman learn implementapplies fixes one diff at a time, only on your yescaveman statscaveman browse <url>/caveman-compress <file>CLAUDE.md, keeps a backupCommands
In your terminal:
In your agent:
/caveman·/ultracave·/megacave/caveman statusshows the mode,/caveman offstops it/caveman-commit/caveman-reviewL42: 🔴 null deref. Guard it./caveman-compress <file>/caveman-stats/caveman-helpcavecrewinvestigate-first·lean-build·surgical-patch·safe-refactor·migration·verify-and-stopcaveman-setup·caveman-discover·caveman-learn·caveman-manage·caveman-optimize·caveman-explore·caveman-evidence-reviewCaveman Cloud
Everything above is for the coding agent you use. Caveman Cloud is for the agents you build.
You ship an agent. Every run costs money. Is it good? Is it cheap? Which change made it worse? Cloud answers.
Sign up free · Try the demo · Book a call · Docs
In the wild
ThePrimeagen: “No way this actually works.” Watch
Adobe Research: CAVEWOMAN, arXiv, June 2026
JetBrains: “It is fun, and it costs you nothing measurable in quality.” Read
Elastic: elastic-caveman on Elasticsearch Labs, April 2026
Hacker News: #1, 904 points
GitHub Trending #1 overall, July 2026 · Trendshift #1 repo of the day, April 2026 · Product Hunt #8 of the day
Started as a joke in April 2026. Now past 100,000 stars. Joke got serious. Voice did not.
Privacy
/caveman-compresssends the file you compress to your configured model provider.caveman telemetry offorDO_NOT_TRACK=1.caveman off routingstops it.caveman providers add, on your own key. Only a request routed to a Caveman Cloud model passes through Caveman Cloud, whole (system prompt, tool results, code), on your Caveman project’s stored key, billed to that project.caveman providers cloud offstops that.The full list, and how to delete sent telemetry: SECURITY.md.
License
Apache-2.0, whole repo. Read it, fork it, ship it, host it. Free like mammoth on open plain. Older releases and third-party notices: LICENSING.md. “Caveman” and the rock logo are trademarks of Julius Brussee.
Cite
🪨 Caveman save you token. Star cost zero. Fair trade.
Docs · Install matrix · Honest numbers · Contributing · Caveman Cloud · Issues