# AI use cases for Caveman (JuliusBrussee/caveman) Repo: [JuliusBrussee/caveman](https://github.com/JuliusBrussee/caveman) | Use cases: 3 | Page: https://dedupedai.com/use-cases/juliusbrussee-caveman/ Caveman is a skill and proxy for coding agents that cuts tokens by making the agent talk tersely and read less, while code stays exact. The README reports 33.2% fewer input tokens across 54 Claude Code runs and links an Adobe Research paper on caveman-style output; it also links a page listing every case where caveman loses. It installs as a skill, a Claude Code plugin, a Gemini CLI extension or an all-agents installer, with an optional free Auto model router. This file holds every use case of this repo. Each one starts with a "## " heading and ends with its URL. - Cut your coding agent's token use with one skill: https://dedupedai.com/use-cases/juliusbrussee-caveman/cut-coding-agent-tokens-with-the-caveman-skill/ - Shrink what your agent reads: logs, test output and web pages: https://dedupedai.com/use-cases/juliusbrussee-caveman/shrink-the-context-your-agent-reads/ - Add Caveman to Gemini CLI as an extension: https://dedupedai.com/use-cases/juliusbrussee-caveman/install-caveman-in-gemini-cli/ ## How to cut your coding agent's token use with one skill with Caveman Repo: [JuliusBrussee/caveman](https://github.com/JuliusBrussee/caveman) | Difficulty: Beginner | Time: 5 minutes | Basis: From the README Install the Caveman skill so the agent answers in short, exact sentences. **Who it is for:** A developer whose agent writes three paragraphs where one line would do. **The problem:** Verbose agent replies burn tokens and bury the answer. **The outcome:** Shorter replies with the same code, switched on and off by command. ### Steps 1. Install for every agent with `npx skills add JuliusBrussee/caveman -g` 2. Or in Claude Code: `/plugin marketplace add JuliusBrussee/caveman` then `/plugin install caveman@caveman` 3. Type `/caveman` (`$caveman` in Codex) if it does not start on its own; `/ultracave` is even shorter 4. Say `stop caveman` to go back to normal ### Prompt to paste into your agent ```text Install JuliusBrussee/caveman as a skill for my coding agent, run the same task once with and once without it, and compare output tokens and whether the code changes were identical. ``` ### Caveat If you pay per request rather than per token (for example GitHub Copilot), shorter answers save nothing, as the README notes. Disclaimer: Third-party repos we don't own or vouch for. AI-generated use cases can be wrong: review everything and run at your own risk. Full text: https://dedupedai.com/disclaimer/ URL: https://dedupedai.com/use-cases/juliusbrussee-caveman/cut-coding-agent-tokens-with-the-caveman-skill/ ## How to shrink what your agent reads: logs, test output and web pages with Caveman Repo: [JuliusBrussee/caveman](https://github.com/JuliusBrussee/caveman) | Difficulty: Beginner | Time: 15 minutes | Basis: From the README Turn on the read-less rocks so tool output and pages come back compressed before they hit the context. **Who it is for:** Someone whose agent context fills up with test logs and HTML. **The problem:** Long command output and pages eat the context window and push out what matters. **The outcome:** Fewer input tokens per session, with `caveman status` showing what is active. ### Steps 1. Install with `npx @caveman-ai/cli@latest` 2. Check what is on with `caveman status` 3. Turn a feature off if it gets in the way, for example `caveman off browse` 4. Compare session token counts before and after ### Prompt to paste into your agent ```text Set up JuliusBrussee/caveman's CLI, show me caveman status, and measure input tokens for one full test-fix loop on this repo with and without it. ``` ### Caveat Disclaimer: Third-party repos we don't own or vouch for. AI-generated use cases can be wrong: review everything and run at your own risk. Full text: https://dedupedai.com/disclaimer/ URL: https://dedupedai.com/use-cases/juliusbrussee-caveman/shrink-the-context-your-agent-reads/ ## How to add Caveman to Gemini CLI as an extension with Caveman Repo: [JuliusBrussee/caveman](https://github.com/JuliusBrussee/caveman) | Difficulty: Beginner | Time: 5 minutes | Basis: From the README Install the Gemini CLI extension so the same terse mode works there. **Who it is for:** A Gemini CLI user who wants the same savings as Claude Code users. **The problem:** Each agent needs its own install path for skills. **The outcome:** Caveman mode available inside Gemini CLI. ### Steps 1. Run `gemini extensions install https://github.com/JuliusBrussee/caveman` 2. Start gemini and type `/caveman` 3. Say `stop caveman` to turn it off ### Prompt to paste into your agent ```text Install JuliusBrussee/caveman as a Gemini CLI extension, confirm /caveman works, and tell me how to uninstall it. ``` ### Caveat Disclaimer: Third-party repos we don't own or vouch for. AI-generated use cases can be wrong: review everything and run at your own risk. Full text: https://dedupedai.com/disclaimer/ URL: https://dedupedai.com/use-cases/juliusbrussee-caveman/install-caveman-in-gemini-cli/