JuliusBrussee/cavemanPublic

why use many token when few token do trick. Skill + local proxy for coding agents: shorter answers, smaller tool output, every original recoverable.

AI summary: An agent skill and local proxy that reduces LLM token consumption by condensing context and output into concise 'cave...

Stars
121.8K
+15 today
Forks
7.1K
Watchers
259
Open issues
61
Open PRs
29
Contributors
~118
Commits
2K
Branches
121
Views (30d)
4

GoApache-2.0Created Apr 4, 2026Last push todayLatest release v4.1.0+1.8K stars this week+7.4K this month

Quick answers

What is caveman?
An agent skill and local proxy that reduces LLM token consumption by condensing context and output into concise 'cave...
What does caveman do?
Caveman is a prompt engineering tool designed to drastically reduce the number of tokens consumed by coding agents like Claude Code, Cursor, and Codex. It operates through two distinct components: a local proxy that actively shrinks the context (like tool schemas, files, and history) sent to the provider, and an agent skill that forces the LLM to output highly condensed, 'caveman-like' responses while preserving byte-exact code and commands. By eliminating conversational filler and aggressively truncating context, it achieves a measured 33.2% reduction in provider-reported input tokens. This approach maintains the agent's core reasoning capabilities and factual accuracy while significantly lowering API costs and latency.
Who is caveman for?
Power users of AI coding assistants and developers managing high API costs. It is targeted at users who prefer direct, unembellished answers and want to optimize their token usage across tools like Claude Code or Cursor.
How do I get started with caveman?
npm install -g @caveman-ai/cli && caveman setup --install
How popular is caveman on GitHub?
JuliusBrussee/caveman has 121,800 stars and 7,104 forks on GitHub, and gained 1,761 stars in the last 7 days.
What license does caveman use?
JuliusBrussee/caveman is released under the Apache-2.0 license.

Star history

since Jul 28, 2026
050K100KJul 2026Sep 2026Oct 2026Nov 2026
121.8K stars as of Nov 21, 2026. Measured daily since Jul 28, 2026; GitHub no longer exposes earlier star timestamps.

Update history

3 recorded
  • Nov 17, 2026Description changed from "🪨 why use many token when few token do trick. Viral skill + proxy for coding agents that cuts 65% of tokens by talking " to "why use many token when few token do trick. Skill + local proxy for coding agents: shorter answers, smaller tool output,".
  • Nov 16, 2026New release v4.1.0 (previously v4.0.1).
  • Oct 13, 2026New release v4.0.1 (previously v3.0.0).

Contribution activity

commits per day, last 52 weeks
NovDecJanFebMarAprMayJunJulAugSepOctNovMonWedFri2025-11-23: 0 commits2025-11-24: 0 commits2025-11-25: 0 commits2025-11-26: 0 commits2025-11-27: 0 commits2025-11-28: 0 commits2025-11-29: 0 commits2025-11-30: 0 commits2025-12-01: 0 commits2025-12-02: 0 commits2025-12-03: 0 commits2025-12-04: 0 commits2025-12-05: 0 commits2025-12-06: 0 commits2025-12-07: 0 commits2025-12-08: 0 commits2025-12-09: 0 commits2025-12-10: 0 commits2025-12-11: 0 commits2025-12-12: 0 commits2025-12-13: 0 commits2025-12-14: 0 commits2025-12-15: 0 commits2025-12-16: 0 commits2025-12-17: 0 commits2025-12-18: 0 commits2025-12-19: 0 commits2025-12-20: 0 commits2025-12-21: 0 commits2025-12-22: 0 commits2025-12-23: 0 commits2025-12-24: 0 commits2025-12-25: 0 commits2025-12-26: 0 commits2025-12-27: 0 commits2025-12-28: 0 commits2025-12-29: 0 commits2025-12-30: 0 commits2025-12-31: 0 commits2026-01-01: 0 commits2026-01-02: 0 commits2026-01-03: 0 commits2026-01-04: 0 commits2026-01-05: 0 commits2026-01-06: 0 commits2026-01-07: 0 commits2026-01-08: 0 commits2026-01-09: 0 commits2026-01-10: 0 commits2026-01-11: 0 commits2026-01-12: 0 commits2026-01-13: 0 commits2026-01-14: 0 commits2026-01-15: 0 commits2026-01-16: 0 commits2026-01-17: 0 commits2026-01-18: 0 commits2026-01-19: 0 commits2026-01-20: 0 commits2026-01-21: 0 commits2026-01-22: 0 commits2026-01-23: 0 commits2026-01-24: 0 commits2026-01-25: 0 commits2026-01-26: 0 commits2026-01-27: 0 commits2026-01-28: 0 commits2026-01-29: 0 commits2026-01-30: 0 commits2026-01-31: 0 commits2026-02-01: 0 commits2026-02-02: 0 commits2026-02-03: 0 commits2026-02-04: 0 commits2026-02-05: 0 commits2026-02-06: 0 commits2026-02-07: 0 commits2026-02-08: 0 commits2026-02-09: 0 commits2026-02-10: 0 commits2026-02-11: 0 commits2026-02-12: 0 commits2026-02-13: 0 commits2026-02-14: 0 commits2026-02-15: 0 commits2026-02-16: 0 commits2026-02-17: 0 commits2026-02-18: 0 commits2026-02-19: 0 commits2026-02-20: 0 commits2026-02-21: 0 commits2026-02-22: 0 commits2026-02-23: 0 commits2026-02-24: 0 commits2026-02-25: 0 commits2026-02-26: 0 commits2026-02-27: 0 commits2026-02-28: 0 commits2026-03-01: 0 commits2026-03-02: 0 commits2026-03-03: 0 commits2026-03-04: 0 commits2026-03-05: 0 commits2026-03-06: 0 commits2026-03-07: 0 commits2026-03-08: 0 commits2026-03-09: 0 commits2026-03-10: 0 commits2026-03-11: 0 commits2026-03-12: 0 commits2026-03-13: 0 commits2026-03-14: 0 commits2026-03-15: 0 commits2026-03-16: 0 commits2026-03-17: 0 commits2026-03-18: 0 commits2026-03-19: 0 commits2026-03-20: 0 commits2026-03-21: 0 commits2026-03-22: 0 commits2026-03-23: 0 commits2026-03-24: 0 commits2026-03-25: 0 commits2026-03-26: 0 commits2026-03-27: 0 commits2026-03-28: 0 commits2026-03-29: 0 commits2026-03-30: 0 commits2026-03-31: 0 commits2026-04-01: 0 commits2026-04-02: 0 commits2026-04-03: 0 commits2026-04-04: 6 commits2026-04-05: 10 commits2026-04-06: 9 commits2026-04-07: 9 commits2026-04-08: 16 commits2026-04-09: 18 commits2026-04-10: 3 commits2026-04-11: 16 commits2026-04-12: 7 commits2026-04-13: 1 commit2026-04-14: 6 commits2026-04-15: 5 commits2026-04-16: 0 commits2026-04-17: 0 commits2026-04-18: 1 commit2026-04-19: 0 commits2026-04-20: 0 commits2026-04-21: 0 commits2026-04-22: 0 commits2026-04-23: 0 commits2026-04-24: 0 commits2026-04-25: 0 commits2026-04-26: 0 commits2026-04-27: 0 commits2026-04-28: 0 commits2026-04-29: 0 commits2026-04-30: 1 commit2026-05-01: 12 commits2026-05-02: 0 commits2026-05-03: 0 commits2026-05-04: 0 commits2026-05-05: 0 commits2026-05-06: 0 commits2026-05-07: 0 commits2026-05-08: 0 commits2026-05-09: 0 commits2026-05-10: 17 commits2026-05-11: 0 commits2026-05-12: 3 commits2026-05-13: 0 commits2026-05-14: 0 commits2026-05-15: 0 commits2026-05-16: 0 commits2026-05-17: 0 commits2026-05-18: 1 commit2026-05-19: 0 commits2026-05-20: 2 commits2026-05-21: 0 commits2026-05-22: 0 commits2026-05-23: 0 commits2026-05-24: 0 commits2026-05-25: 0 commits2026-05-26: 0 commits2026-05-27: 0 commits2026-05-28: 0 commits2026-05-29: 0 commits2026-05-30: 0 commits2026-05-31: 0 commits2026-06-01: 9 commits2026-06-02: 0 commits2026-06-03: 0 commits2026-06-04: 0 commits2026-06-05: 0 commits2026-06-06: 0 commits2026-06-07: 1 commit2026-06-08: 0 commits2026-06-09: 0 commits2026-06-10: 1 commit2026-06-11: 0 commits2026-06-12: 7 commits2026-06-13: 0 commits2026-06-14: 5 commits2026-06-15: 0 commits2026-06-16: 0 commits2026-06-17: 0 commits2026-06-18: 0 commits2026-06-19: 0 commits2026-06-20: 0 commits2026-06-21: 0 commits2026-06-22: 0 commits2026-06-23: 0 commits2026-06-24: 1 commit2026-06-25: 0 commits2026-06-26: 0 commits2026-06-27: 0 commits2026-06-28: 0 commits2026-06-29: 0 commits2026-06-30: 1 commit2026-07-01: 8 commits2026-07-02: 14 commits2026-07-03: 5 commits2026-07-04: 0 commits2026-07-05: 0 commits2026-07-06: 0 commits2026-07-07: 0 commits2026-07-08: 0 commits2026-07-09: 0 commits2026-07-10: 0 commits2026-07-11: 0 commits2026-07-12: 0 commits2026-07-13: 0 commits2026-07-14: 0 commits2026-07-15: 0 commits2026-07-16: 0 commits2026-07-17: 0 commits2026-07-18: 0 commits2026-07-19: 0 commits2026-07-20: 0 commits2026-07-21: 6 commits2026-07-22: 0 commits2026-07-23: 1 commit2026-07-24: 0 commits2026-07-25: 0 commits2026-07-26: 0 commits2026-07-27: 1 commit2026-07-28: 1 commit2026-07-29: 0 commits2026-07-30: 0 commits2026-07-31: 0 commits2026-08-01: 3 commits2026-08-02: 2 commits2026-08-03: 4 commits2026-08-04: 1 commit2026-08-05: 2 commits2026-08-06: 0 commits2026-08-07: 0 commits2026-08-08: 4 commits2026-08-09: 1 commit2026-08-10: 1 commit2026-08-11: 6 commits2026-08-12: 6 commits2026-08-13: 7 commits2026-08-14: 1 commit2026-08-15: 9 commits2026-08-16: 8 commits2026-08-17: 1 commit2026-08-18: 11 commits2026-08-19: 39 commits2026-08-20: 16 commits2026-08-21: 6 commits2026-08-22: 0 commits2026-08-23: 27 commits2026-08-24: 1 commit2026-08-25: 11 commits2026-08-26: 1 commit2026-08-27: 0 commits2026-08-28: 0 commits2026-08-29: 13 commits2026-08-30: 53 commits2026-08-31: 0 commits2026-09-01: 2 commits2026-09-02: 14 commits2026-09-03: 22 commits2026-09-04: 5 commits2026-09-05: 4 commits2026-09-06: 1 commit2026-09-07: 5 commits2026-09-08: 42 commits2026-09-09: 2 commits2026-09-10: 7 commits2026-09-11: 4 commits2026-09-12: 1 commit2026-09-13: 7 commits2026-09-14: 45 commits2026-09-15: 10 commits2026-09-16: 13 commits2026-09-17: 5 commits2026-09-18: 6 commits2026-09-19: 4 commits2026-09-20: 6 commits2026-09-21: 7 commits2026-09-22: 7 commits2026-09-23: 8 commits2026-09-24: 25 commits2026-09-25: 11 commits2026-09-26: 8 commits2026-09-27: 36 commits2026-09-28: 26 commits2026-09-29: 7 commits2026-09-30: 16 commits2026-10-01: 8 commits2026-10-02: 49 commits2026-10-03: 14 commits2026-10-04: 9 commits2026-10-05: 202 commits2026-10-06: 125 commits2026-10-07: 35 commits2026-10-08: 47 commits2026-10-09: 23 commits2026-10-10: 187 commits2026-10-11: 13 commits2026-10-12: 16 commits2026-10-13: 14 commits2026-10-14: 16 commits2026-10-15: 5 commits2026-10-16: 20 commits2026-10-17: 11 commits2026-10-18: 6 commits2026-10-19: 7 commits2026-10-20: 1 commit2026-10-21: 10 commits2026-10-22: 8 commits2026-10-23: 17 commits2026-10-24: 8 commits2026-10-25: 13 commits2026-10-26: 5 commits2026-10-27: 4 commits2026-10-28: 3 commits2026-10-29: 5 commits2026-10-30: 4 commits2026-10-31: 5 commits2026-11-01: 2 commits2026-11-02: 13 commits2026-11-03: 7 commits2026-11-04: 6 commits2026-11-05: 4 commits2026-11-06: 37 commits2026-11-07: 1 commit2026-11-08: 3 commits2026-11-09: 0 commits2026-11-10: 0 commits2026-11-11: 1 commit2026-11-12: 5 commits2026-11-13: 2 commits2026-11-14: 0 commits2026-11-15: 0 commits2026-11-16: 0 commits2026-11-17: 0 commits2026-11-18: 0 commits2026-11-19: 0 commits2026-11-20: 0 commits2026-11-21: 0 commits
1,765 commits in the last yearLessMore

Signals and awards

derived from tracked data
  • Landmark project

    121,800 stars

  • Very active

    1,765 commits in 52 weeks

  • Community-driven

    ~118 contributors

  • Well documented

    High community health score

  • Permissive license

    Apache-2.0

  • Continuous integration

    Automated checks passing

  • Repeat trending

    39 trending appearances

  • Top 10% tracked

    Rank 62 of 1243

What caveman does

Caveman is a prompt engineering tool designed to drastically reduce the number of tokens consumed by coding agents like Claude Code, Cursor, and Codex. It operates through two distinct components: a local proxy that actively shrinks the context (like tool schemas, files, and history) sent to the provider, and an agent skill that forces the LLM to output highly condensed, 'caveman-like' responses while preserving byte-exact code and commands. By eliminating conversational filler and aggressively truncating context, it achieves a measured 33.2% reduction in provider-reported input tokens. This approach maintains the agent's core reasoning capabilities and factual accuracy while significantly lowering API costs and latency.

Power users of AI coding assistants and developers managing high API costs. It is targeted at users who prefer direct, unembellished answers and want to optimize their token usage across tools like Claude Code or Cursor.

  • Context Shrinking Proxy: Intercepts and compresses tool schemas, files, and conversation history before they reach the provider API.
  • Concise Output Generation: Forces the agent to eliminate conversational boilerplate and answer using extreme brevity.
  • Byte-Exact Code Preservation: Ensures that while natural language is compressed, all generated code, terminal commands, and system errors remain perfectly intact.
  • Broad Agent Compatibility: Works natively across more than 30 different coding agents, including Claude Code, Gemini CLI, and OpenCode.
  • Flexible Deployment: Offers both a local proxy for input compression and a direct agent skill for output reduction, which can be used independently or together.

Where teams use it

Cost Reduction for AI Coding

Helps developers minimize API billing costs by stripping unnecessary tokens from both the input context and the model's responses during continuous coding sessions.

Latency Optimization

Accelerates agent response times by reducing the total volume of text the LLM needs to process and generate, focusing compute entirely on the actionable code.

Streamlined Terminal Output

Cleans up the developer's terminal interface by preventing the agent from generating long, conversational explanations for simple code fixes.

Maximized Context Windows

Allows developers to fit larger codebases or more extensive chat histories into a model's context window by aggressively compressing the non-essential text.

Getting started: npm install -g @caveman-ai/cli && caveman setup --install

README

main branch

New: Caveman Code. Free.

The most efficient coding agent. Same model, same work, smaller bill.

21% cheaper

than Codex
Terminal-Bench 4, per solved task

14% cheaper

than Claude Code
Terminal-Bench 4, 97 vs 83 solved

76% cheaper

than Claude Code
full workday, 4 connected apps

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


Caveman

why many token when few do trick

Caveman make your AI agent say less and read less. Code stay exact. Brain still big.

GitHub stars npm downloads 30+ agents License

33.2% fewer

input tokens, whole sessions
54 Claude Code runs, 18 of 18 answers right

54% cheaper

with Auto, same score
85 coding tasks vs Opus 5.5 at xhigh, interim

397 → 76

tokens for one test run
Blocks vs | tail -20

1.4 to 2.4× cheaper

with caveman-style output
Adobe 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


Normal agent · 63 tokens Caveman agent · 20 tokens

The reason your React component is re-rendering is likely because you're creating a new object reference on each render cycle. When you pass an inline object as a prop, React's shallow comparison sees it as a different object every time, which triggers a re-render. I'd recommend using useMemo to memoize the object.

New object ref each render, so React re-renders. Wrap the prop in useMemo.

Same fix. 63 token become 20. Brain still big.

Pick your club:

Skill Same answer Tokens
/caveman New object ref each render, so React re-renders. Wrap the prop in useMemo. 20
/ultracave Inline object prop, new ref, re-render. useMemo. 14
/megacave 新參照致重繪。useMemo。 13

Token counts: tiktoken o200k.

Install

npx @caveman-ai/cli@latest

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 claude stays 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 setup opens 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.

Rock What it does Measured
Say less Agent answers short. Code stays exact 63 → 20 tokens, same fix
Read less Logs, JSON, CSV, test output shrink before the model reads them 33.2% fewer input tokens per session
Auto · new Picks the model and the effort for each ask 54% cheaper than Opus at xhigh, same score
Blocks · new Code mode: agent keeps the scripts it writes and runs them again 397 → 76 tokens for one test run
Waste fixes Finds where your tokens go and fixes it. New: keeps the prompt cache warm while you pause Not measured yet
Browse Web pages come back compressed 129.8× smaller page

caveman status shows what is on. caveman off browse turns one off. Only Auto needs an account, and it is free. 4.0.0 release notes

Proof

Outside labs first. Ours come with each rock below. Nothing rounded up. Red rows stay red.

Who Setup Result
Adobe Research, CAVEWOMAN paper (cites this repo) Caveman-style output, eight models, five datasets Cost cut 1.4 to 2.4× per model, up to 3×
Elastic, Elasticsearch Labs Caveman mode remade for Elasticsearch, eight live MCP scenarios 63.6% fewer response tokens. "Zero information loss."
JetBrains 86 real coding tasks, paired A/B No measurable quality loss (p = 0.82). 8.5% fewer output tokens

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.

Rule What it means
Answer first [thing] [action] [reason]. [next step]. No greeting, no "let me", no recap
One idea per sentence Built on ASD-STE100, the controlled English of aircraft maintenance manuals: 20 words max, active voice, one term per thing
Meaning never dropped Articles can go. not, never, no, only never go. Numbers and units stay exact
Payload verbatim Code, commands, paths, and error messages untouched, character for character
No theater No "me think". No chatter between tool calls. If caveman phrasing isn't shorter, plain wins
Knows when to stop Security warnings, irreversible actions, and confused users get full sentences. Then grunt resumes
Your prompts stay yours Never rewritten. Research say that backfire

Ten dev questions on claude-opus-5-5:

Instruction Output tokens
None 6,983
Answer concisely. 4,334
/caveman 4,119
/ultracave 2,693

"Be concise" is the real baseline; new models already know it. On top of it, /caveman cuts 3% more at the median and /ultracave 35% 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 loses

Only want this rock? The skill alone, no proxy:

npx skills add JuliusBrussee/caveman -g

skills.sh

Type /caveman ($caveman in Codex) if it doesn't start on its own. Say stop caveman to 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.

# Claude Code plugin, auto-starts every interactive session, subagents too
claude plugin marketplace add JuliusBrussee/caveman && claude plugin install caveman@caveman

# Gemini CLI
gemini extensions install https://github.com/JuliusBrussee/caveman

# Every agent on your machine at once, plus the Claude Code statusline badge (Node.js 22.13+)
curl -fsSL https://raw.githubusercontent.com/JuliusBrussee/caveman/v4.1.0/install.sh | bash

Already inside Claude Code? Same rock, slash form. Type these where you talk to Claude:

/plugin marketplace add JuliusBrussee/caveman
/plugin install caveman@caveman

Windows, PowerShell 5.1+:

irm https://raw.githubusercontent.com/JuliusBrussee/caveman/v4.1.0/install.ps1 | iex

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#v4.1.0

Changed your mind: npx -y github:JuliusBrussee/caveman -- --uninstall (on npm 12+, add --allow-git=root too)

Only grabbed the skill? npx skills remove JuliusBrussee/caveman -g take it back out.

Install broke? Open your agent in this repo and say "Read CLAUDE.md and 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.

File type The file, through caveman File saved Whole Claude Code session, 3 runs Session saved
CSV 28,041 → 314 98.9% 165,823 → 74,484 55.1%
Logs 22,810 → 348 98.5% 148,807 → 74,068 50.2%
YAML 20,447 → 178 99.1% 132,124 → 71,027 46.2%
Test output 18,806 → 203 98.9% 150,377 → 108,514 27.8%
JSON 18,837 → 281 98.5% 147,975 → 108,939 26.4%
HTML 21,670 → 21,670 none yet 140,687 → 154,641 9.9% worse
All six 130,611 → 22,994 82.4% 885,793 → 591,673 33.2%

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. Method

Every original stays on your disk. The agent can pull it back any time.

Auto: 54% cheaper, same score

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.

Setup Tasks solved Cost per try Median time
Opus 5.5, xhigh effort, every ask 44.3% $7.83 35.2 min
Auto, Claude models only 43.7% $3.59 15.9 min
Auto, plus OpenAI and open models 42.9% $3.47 16.7 min

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

caveman login                  # free account. Auto shows up in the model picker
caveman providers add openai   # optional: give Auto more models, on your own key

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.

Input Raw What agents type now Block
go test -v ./..., 326 tests 10,561 397 · | tail -20 76 · test-summary
1.4 MB JSON file 251,893 236 · their own heredoc 227 · json-peek
366-line Go file 3,111 250 · grep -n 489 · grep-defs

Tokens 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-defs loses to a bare grep; row stays. Method

Comes 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.

caveman-blocks scan    # what your agents retyped this month. Script text never leaves your machine
caveman-blocks init    # rules and block index into this repo's AGENTS.md

Pre-release (v0.1.0-rc.5). Python only, no sandbox yet. Own repo, own stars: caveman-ai/blocks

Smaller rocks

Rock What it does Measured
caveman learn Ranks where your tokens go, from agent history already on disk. caveman learn implement applies fixes one diff at a time, only on your yes Not yet
Warm cache · new Step away and the provider forgets your session, so the next ask pays full price. Caveman keeps Anthropic's prompt cache warm while you pause, only when your own history says you come back in time. Every warm is listed in caveman stats Not yet
caveman browse <url> A compressed web page for the agent 121 tokens instead of 15,704 for a Playwright snapshot of a 200-row table. Tiny forms lose 2.3×. Bench
/caveman-compress <file> Shrinks memory files like CLAUDE.md, keeps a backup 4,138 tokens instead of 6,198 across five files, 22.8% to 49.1% each, every heading, code block, and path intact. Bench
Middleware Same shrinking inside the Vercel AI SDK, LangChain, OpenAI, or Anthropic call you already make. TypeScript · Python · Every framework · One container for the team Not yet

Caveman Learn report: a summary and savings cards on the left; the biggest places tokens go, with one fix opened, and a chart of how full sessions get on the right

Commands

In your terminal:

caveman claude                # start an agent on caveman: codex · gemini · opencode · aider · kilo · qwen · hermes · openclaw · pi
caveman status                # what is on, which agents it reaches, one next step
caveman on browse             # switch one rock on or off: output · input · waste-fixes · routing · scripts · browse
caveman login                 # free account, turns on Auto
caveman learn                 # where your tokens go
caveman trial -- claude       # A/B a real session on your own work
caveman shrink -- pnpm test   # compress noisy command output
caveman convert --dry-run     # pixel mode: skills the model reads as images
caveman stats                 # your token history
caveman doctor                # one problem per line, with its fix

In your agent:

Command What it does
/caveman · /ultracave · /megacave The voice, the grunt, the 文言文. /caveman status shows the mode, /caveman off stops it
/caveman-commit One-line Conventional Commit
/caveman-review One finding per line: L42: 🔴 null deref. Guard it.
/caveman-compress <file> Shrinks memory files and backs up the original
/caveman-stats Real token usage for this Claude Code session
/caveman-help Every mode and command on one screen
cavecrew Subagents that find, edit, and review code, then report back in caveman
investigate-first · lean-build · surgical-patch · safe-refactor · migration · verify-and-stop Work patterns that write less code. Your agent picks them up when a task fits
caveman-setup · caveman-discover · caveman-learn · caveman-manage · caveman-optimize · caveman-explore · caveman-evidence-review Drive the proxy from inside your agent

Caveman 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.

Cloud does What you get
Observability Every run traced to the work and what it cost. Sets itself up from your repo, your keys, and the logs or OpenTelemetry you already have
Evals Grades how good the answers are, on your own tasks, from your real traffic
Autopilot Proposes a cheaper change, tests it on your recorded work, ships only when the evals say answers are still as good. You approve the pull request
Proof Every number says if it is measured, inferred, or verified. A saving counts only after real traffic shows it

Sign up free  ·  Try the demo  ·  Book a call  ·  Docs

In the wild

ThePrimeagen reacts to Caveman: No way this actually works

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.


Star History Chart

Privacy

  • The skill runs on your machine and sends nothing. One exception: /caveman-compress sends the file you compress to your configured model provider.
  • The CLI sends usage stats by default: commands run, token counts, caveman-blocks commands and usage counts, a random install ID, OS, and IP. Never your prompts, code, file paths, or scripts you wrote. One person maintains this for free, and those stats show what to build next. Off for good, Blocks included: caveman telemetry off or DO_NOT_TRACK=1.
  • Auto, only once you sign in and pick it: each ask sends Caveman Cloud your latest ask (with whatever your agent attaches to it, such as CLAUDE.md, memory or @-mentioned files), the one before it, and the end of the agent's last reply, plus each request's tool names, effort, a few agent headers, and the previous request's token counts. On the Free plan Caveman may keep it to improve routing. caveman off routing stops it.
  • Where requests go: from your machine to your provider, or to another provider you added with 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 off stops 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

@software{brussee2026caveman,
  author = {Brussee, Julius},
  title  = {Caveman: why many token when few do trick},
  year   = {2026},
  url    = {https://github.com/JuliusBrussee/caveman}
}

🪨 Caveman save you token. Star cost zero. Fair trade.

Docs · Install matrix · Honest numbers · Contributing · Caveman Cloud · Issues
View on GitHub

Recent activity

commits and pull requests

Releases and announcements

56 total
  1. Caveman 4.1.0v4.1.0Nov 8, 20261.9K downloads

    Caveman 4.1.0 is a maintenance release: bug fixes across compress, OpenCode, Hermes, Pi, the proxy and the hooks, a recovery store that no longer fills up and stops compressing, and a few small switches people asked for. caveman-compress now refuses a file with a key or token in it, and installs from skill hubs again. ### Action needed - **`forward_headers` under a `providers.<name>` block now stops the proxy from starting.** Before, it was silently ignored: no headers were forwarded and the only symptom was an upstream 401. Configure that upstream as a compat mount instead: `compat.<name>` with `base_url` and `forward_headers`, plus `wire_dialect: anthropic` if it speaks the Anthropic protocol (see `docs/technical/configuration.md`) (#1458). - **caveman-compress refuses a file whose text holds a credential**, such as a private key block or a cloud, Git host or model API token. Compression sends the file to your model provider, so move the secret into an env var or secret manager first. The error names the line, never the secret. Placeholders like `sk-...` or `$OPENAI_API_KEY` still compress (#1471). ### Changed - **Old recovery handles can now expire.** The recovery store used

  2. @caveman-ai/middleware 1.1.0middleware-ts-v1.1.0Nov 8, 20266 downloads

    - Security: the MCP adapter now requires `@modelcontextprotocol/sdk` 1.32.0 or later. Earlier releases followed redirects to any origin and handed it custom headers, and on 307/308 the request body including OAuth tokens ([GHSA-6prh-2h8m-c8cw](https://osv.dev/vulnerability/GHSA-6prh-2h8m-c8cw)). With 1.30.x or 1.31.x installed, MCP results pass through uncompressed and a warning names the advisory and the version to upgrade to. - `@anthropic-ai/sdk` 0.132.0 through 0.137.0 and the 1.x major (tested 1.0.0 and 1.1.0) are inside the supported range (`>=0.124.0 <2`). Before this, the latest SDK passed content through with an `unsupported_version` warning. - Mastra: call options forward again across `@mastra/core`'s vendored `@ai-sdk/provider` types, which lagged the real package and failed `tsc`. ### Verify this release - Registry provenance (trusted publishing from this workflow): https://www.npmjs.com/package/@caveman-ai/middleware/v/1.1.0#provenance - CycloneDX SBOM of the published dependency graph: attached `.cdx.json` - Built from annotated, GitHub-verified tag `middleware-ts-v1.1.0` on `main`

  3. caveman-middleware 1.0.1middleware-python-v1.0.1Nov 8, 20266 downloads

    - Security: the `litellm` extra excludes litellm 1.107.0, which upstream reported compromised ([security incident, 2026-10-21](https://docs.litellm.ai/blog/security-incident-oct-21-2026)), and 1.106.0 to 1.106.2, which import the optional `mcp` package on ordinary tool calls. The extra also requires a patched `filelock` (4.2.1 or later). - `anthropic` 2.x is inside the supported range (`>=1.0,<3`). Before this, the latest SDK passed content through with an `unsupported_version` warning. ### Verify this release - Registry provenance (trusted publishing from this workflow): https://pypi.org/project/caveman-middleware/1.0.1/#files - CycloneDX SBOM of the published dependency graph: attached `.cdx.json` - Built from annotated, GitHub-verified tag `middleware-python-v1.0.1` on `main`

  4. bin-v4.1.0bin-v4.1.0Nov 7, 2026123.1K downloads

    Signed Caveman runtime companions for macOS, Linux, and Windows. The binaries are licensed under the Apache License 2.0 (LICENSE, NOTICE); see LICENSING.md and the attached third-party notices. The ghcr.io caveman-proxy image is still built without cgo, so its code compressor handles Go source only; the caveman-proxy and caveman-engine binaries here also compress TypeScript, JavaScript, Python, Rust, Java and C/C++.

  5. bin-v4.0.2bin-v4.0.2Oct 23, 202622.6K downloads

    Signed Caveman runtime companions for macOS, Linux, and Windows. The binaries are licensed under the Apache License 2.0 (LICENSE, NOTICE); see LICENSING.md and the attached third-party notices. The ghcr.io caveman-proxy image is still built without cgo, so its code compressor handles Go source only; the caveman-proxy and caveman-engine binaries here also compress TypeScript, JavaScript, Python, Rust, Java and C/C++.

Code frequency

additions and deletions
+267.3K-267.3KWeek of 2026-03-29: +351 linesWeek of 2026-03-29: -25 linesWeek of 2026-04-05: +12,833 linesWeek of 2026-04-05: -5,904 linesWeek of 2026-04-12: +1,382 linesWeek of 2026-04-12: -198 linesWeek of 2026-04-19: +0 linesWeek of 2026-04-19: -0 linesWeek of 2026-04-26: +5,018 linesWeek of 2026-04-26: -810 linesWeek of 2026-05-03: +0 linesWeek of 2026-05-03: -0 linesWeek of 2026-05-10: +8,828 linesWeek of 2026-05-10: -7,464 linesWeek of 2026-05-17: +45 linesWeek of 2026-05-17: -20 linesWeek of 2026-05-24: +0 linesWeek of 2026-05-24: -0 linesWeek of 2026-05-31: +1,395 linesWeek of 2026-05-31: -165 linesWeek of 2026-06-07: +246 linesWeek of 2026-06-07: -378 linesWeek of 2026-06-14: +619 linesWeek of 2026-06-14: -18 linesWeek of 2026-06-21: +23 linesWeek of 2026-06-21: -0 linesWeek of 2026-06-28: +2,106 linesWeek of 2026-06-28: -445 linesWeek of 2026-07-05: +0 linesWeek of 2026-07-05: -0 linesWeek of 2026-07-12: +0 linesWeek of 2026-07-12: -0 linesWeek of 2026-07-19: +4,197 linesWeek of 2026-07-19: -2,590 linesWeek of 2026-07-26: +994 linesWeek of 2026-07-26: -144 linesWeek of 2026-08-02: +548 linesWeek of 2026-08-02: -156 linesWeek of 2026-08-09: +267,280 linesWeek of 2026-08-09: -3,959 linesWeek of 2026-08-16: +24,656 linesWeek of 2026-08-16: -2,717 linesWeek of 2026-08-23: +10,738 linesWeek of 2026-08-23: -12,074 linesWeek of 2026-08-30: +14,075 linesWeek of 2026-08-30: -12,955 linesWeek of 2026-09-06: +20,314 linesWeek of 2026-09-06: -2,906 linesWeek of 2026-09-13: +183,475 linesWeek of 2026-09-13: -156,900 linesWeek of 2026-09-20: +51,019 linesWeek of 2026-09-20: -60,962 linesWeek of 2026-09-27: +21,712 linesWeek of 2026-09-27: -10,122 linesWeek of 2026-10-04: +80,664 linesWeek of 2026-10-04: -19,605 linesWeek of 2026-10-11: +6,874 linesWeek of 2026-10-11: -1,156 linesWeek of 2026-10-18: +3,986 linesWeek of 2026-10-18: -410 linesWeek of 2026-10-25: +3,062 linesWeek of 2026-10-25: -257 linesWeek of 2026-11-01: +10,225 linesWeek of 2026-11-01: -2,053 linesWeek of 2026-11-08: +532 linesWeek of 2026-11-08: -47 linesWeek of 2026-11-15: +0 linesWeek of 2026-11-15: -0 linesMar 29, 2026Nov 15, 2026
+737.2K lines added, -304.4K removed over the last year.

Commits per week

last 52 weeks
6280Week of 2025-11-23: 0 commitsWeek of 2025-11-30: 0 commitsWeek of 2025-12-07: 0 commitsWeek of 2025-12-14: 0 commitsWeek of 2025-12-21: 0 commitsWeek of 2025-12-28: 0 commitsWeek of 2026-01-04: 0 commitsWeek of 2026-01-11: 0 commitsWeek of 2026-01-18: 0 commitsWeek of 2026-01-25: 0 commitsWeek of 2026-02-01: 0 commitsWeek of 2026-02-08: 0 commitsWeek of 2026-02-15: 0 commitsWeek of 2026-02-22: 0 commitsWeek of 2026-03-01: 0 commitsWeek of 2026-03-08: 0 commitsWeek of 2026-03-15: 0 commitsWeek of 2026-03-22: 0 commitsWeek of 2026-03-29: 6 commitsWeek of 2026-04-05: 81 commitsWeek of 2026-04-12: 20 commitsWeek of 2026-04-19: 0 commitsWeek of 2026-04-26: 13 commitsWeek of 2026-05-03: 0 commitsWeek of 2026-05-10: 20 commitsWeek of 2026-05-17: 3 commitsWeek of 2026-05-24: 0 commitsWeek of 2026-05-31: 9 commitsWeek of 2026-06-07: 9 commitsWeek of 2026-06-14: 5 commitsWeek of 2026-06-21: 1 commitsWeek of 2026-06-28: 28 commitsWeek of 2026-07-05: 0 commitsWeek of 2026-07-12: 0 commitsWeek of 2026-07-19: 7 commitsWeek of 2026-07-26: 5 commitsWeek of 2026-08-02: 13 commitsWeek of 2026-08-09: 31 commitsWeek of 2026-08-16: 81 commitsWeek of 2026-08-23: 53 commitsWeek of 2026-08-30: 100 commitsWeek of 2026-09-06: 62 commitsWeek of 2026-09-13: 90 commitsWeek of 2026-09-20: 72 commitsWeek of 2026-09-27: 156 commitsWeek of 2026-10-04: 628 commitsWeek of 2026-10-11: 95 commitsWeek of 2026-10-18: 57 commitsWeek of 2026-10-25: 39 commitsWeek of 2026-11-01: 70 commitsWeek of 2026-11-08: 11 commitsWeek of 2026-11-15: 0 commitsNov 23, 2025Nov 15, 2026
1.8K commits in the last 52 weeks.

When work happens

weekday and hour
SunMonTueWedThuFriSat036912151821Sun 0:00 — 9 commitsSun 1:00 — 15 commitsSun 2:00 — 7 commitsSun 3:00 — 7 commitsSun 4:00 — 1 commitsSun 5:00 — 13 commitsSun 6:00 — 9 commitsSun 7:00 — 26 commitsSun 8:00 — 9 commitsSun 9:00 — 3 commitsSun 10:00 — 23 commitsSun 11:00 — 2 commitsSun 12:00 — 6 commitsSun 13:00 — 18 commitsSun 14:00 — 5 commitsSun 15:00 — 4 commitsSun 16:00 — 2 commitsSun 17:00 — 21 commitsSun 18:00 — 17 commitsSun 19:00 — 7 commitsSun 20:00 — 7 commitsSun 21:00 — 10 commitsSun 22:00 — 6 commitsSun 23:00 — 1 commitsMon 0:00 — 1 commitsMon 1:00 — 2 commitsMon 2:00 — 6 commitsMon 3:00 — 1 commitsMon 4:00 — 0 commitsMon 5:00 — 0 commitsMon 6:00 — 1 commitsMon 7:00 — 37 commitsMon 8:00 — 9 commitsMon 9:00 — 14 commitsMon 10:00 — 5 commitsMon 11:00 — 4 commitsMon 12:00 — 5 commitsMon 13:00 — 0 commitsMon 14:00 — 1 commitsMon 15:00 — 17 commitsMon 16:00 — 15 commitsMon 17:00 — 27 commitsMon 18:00 — 15 commitsMon 19:00 — 15 commitsMon 20:00 — 25 commitsMon 21:00 — 119 commitsMon 22:00 — 23 commitsMon 23:00 — 12 commitsTue 0:00 — 5 commitsTue 1:00 — 19 commitsTue 2:00 — 19 commitsTue 3:00 — 12 commitsTue 4:00 — 0 commitsTue 5:00 — 0 commitsTue 6:00 — 0 commitsTue 7:00 — 28 commitsTue 8:00 — 0 commitsTue 9:00 — 6 commitsTue 10:00 — 27 commitsTue 11:00 — 34 commitsTue 12:00 — 8 commitsTue 13:00 — 13 commitsTue 14:00 — 13 commitsTue 15:00 — 5 commitsTue 16:00 — 3 commitsTue 17:00 — 5 commitsTue 18:00 — 15 commitsTue 19:00 — 8 commitsTue 20:00 — 14 commitsTue 21:00 — 16 commitsTue 22:00 — 12 commitsTue 23:00 — 12 commitsWed 0:00 — 15 commitsWed 1:00 — 3 commitsWed 2:00 — 0 commitsWed 3:00 — 0 commitsWed 4:00 — 6 commitsWed 5:00 — 2 commitsWed 6:00 — 0 commitsWed 7:00 — 37 commitsWed 8:00 — 5 commitsWed 9:00 — 2 commitsWed 10:00 — 9 commitsWed 11:00 — 10 commitsWed 12:00 — 6 commitsWed 13:00 — 9 commitsWed 14:00 — 12 commitsWed 15:00 — 15 commitsWed 16:00 — 9 commitsWed 17:00 — 7 commitsWed 18:00 — 25 commitsWed 19:00 — 9 commitsWed 20:00 — 4 commitsWed 21:00 — 3 commitsWed 22:00 — 8 commitsWed 23:00 — 9 commitsThu 0:00 — 16 commitsThu 1:00 — 24 commitsThu 2:00 — 20 commitsThu 3:00 — 1 commitsThu 4:00 — 12 commitsThu 5:00 — 3 commitsThu 6:00 — 3 commitsThu 7:00 — 27 commitsThu 8:00 — 7 commitsThu 9:00 — 0 commitsThu 10:00 — 0 commitsThu 11:00 — 4 commitsThu 12:00 — 2 commitsThu 13:00 — 10 commitsThu 14:00 — 8 commitsThu 15:00 — 12 commitsThu 16:00 — 3 commitsThu 17:00 — 12 commitsThu 18:00 — 2 commitsThu 19:00 — 5 commitsThu 20:00 — 13 commitsThu 21:00 — 10 commitsThu 22:00 — 1 commitsThu 23:00 — 6 commitsFri 0:00 — 15 commitsFri 1:00 — 47 commitsFri 2:00 — 13 commitsFri 3:00 — 1 commitsFri 4:00 — 17 commitsFri 5:00 — 4 commitsFri 6:00 — 1 commitsFri 7:00 — 19 commitsFri 8:00 — 1 commitsFri 9:00 — 1 commitsFri 10:00 — 2 commitsFri 11:00 — 4 commitsFri 12:00 — 3 commitsFri 13:00 — 12 commitsFri 14:00 — 16 commitsFri 15:00 — 3 commitsFri 16:00 — 12 commitsFri 17:00 — 12 commitsFri 18:00 — 5 commitsFri 19:00 — 13 commitsFri 20:00 — 5 commitsFri 21:00 — 2 commitsFri 22:00 — 3 commitsFri 23:00 — 1 commitsSat 0:00 — 82 commitsSat 1:00 — 28 commitsSat 2:00 — 20 commitsSat 3:00 — 11 commitsSat 4:00 — 2 commitsSat 5:00 — 22 commitsSat 6:00 — 13 commitsSat 7:00 — 25 commitsSat 8:00 — 7 commitsSat 9:00 — 3 commitsSat 10:00 — 2 commitsSat 11:00 — 7 commitsSat 12:00 — 11 commitsSat 13:00 — 9 commitsSat 14:00 — 1 commitsSat 15:00 — 3 commitsSat 16:00 — 12 commitsSat 17:00 — 6 commitsSat 18:00 — 9 commitsSat 19:00 — 4 commitsSat 20:00 — 2 commitsSat 21:00 — 2 commitsSat 22:00 — 10 commitsSat 23:00 — 6 commits
Commit volume by weekday and hour (UTC). Larger dots mean more commits.

Who is committing

last 52 weeks
Maintainer commits1,484 (73%)
Community commits558 (27%)

2,042 commits in total over the last year.

DateListRankStars gained
Jul 15, 2026daily#14+7
Jul 14, 2026daily#16+5
Jul 13, 2026daily#8+4
Jul 12, 2026daily#12+3
Jul 5, 2026daily#7+31
Jul 4, 2026daily#1+42
Jul 3, 2026daily#2+32
Jul 2, 2026daily#4+43
May 11, 2026daily#24+92
May 7, 2026daily#17+57
May 6, 2026daily#25+57
May 5, 2026daily#25+53
May 1, 2026daily#15+69
Apr 30, 2026daily#15+53
Apr 29, 2026daily#12+90
  • morluto/rea

    Reverse engineer anything with agents, from app behavior down to native binaries.

    471.6K stars · TypeScript

  • openclaw/openclaw

    The AI that really does things. Any OS. Any Platform. The lobster way. 🦞

    394.9K stars · TypeScript

  • obra/superpowers

    An agentic skills framework & software development methodology that works.

    322.9K stars · Shell

  • 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.

    307.9K stars · JavaScript

  • NousResearch/hermes-agent

    The agent that grows with you

    265K stars · Python

  • n8n-io/n8n

    Fair-code workflow automation platform with native AI capabilities. Combine visual building with custom code, self-host or cloud, 400+ integrations.

    211.6K stars · TypeScript