Guide · Agent skill
How to use the Ultracave terse responses agent skill
Updated · written with AI from the project's own documentation and checked against our catalogue data
Ultracave is an agent skill that makes an AI coding agent reply in very short, clipped fragments while keeping code, commands, numbers and error messages exact. This guide explains what the skill does, how to install it from the ZIP or the Claude apps, how to switch it on and off, and where it falls short. You do not need prior experience with agent skills to follow it.
Ultracave Terse ResponsesJuliusBrussee/caveman · Apache-2.0 Download the skill Quick start
- 1Download the ZIP. Get the ultracave ZIP from this skill's page on the directory.
- 2Unzip the skill. Extract it into ~/.claude/skills/ for personal use, or into a project's .claude/skills/ folder.
- 3Or upload in the Claude apps. In the Claude apps, go to Settings → Capabilities → Skills and upload the skill.
- 4Start the mode. Type /ultracave or /caveman ultra in your agent session.
- 5Check the mode. Type /caveman status to see the current mode.
- 6Stop the mode. Say "stop caveman" or "normal mode" to return to normal replies.
What the ultracave skill makes the agent do
Once installed and switched on, the agent answers in very short fragments. It drops articles, connecting words and facts it has already stated, so each fact appears once. Technical content is kept intact: code, commands, paths, API names, error strings, numbers and units are never cut.
The mode stays on for the whole session until you say "stop caveman" or "normal mode". It only changes how the agent writes its replies. It is not a tool that does work by itself, so it will not run code or edit files on its own.
- Fragments instead of full sentences, when the meaning stays clear
- One word when one word is enough
- Each fact stated once, with no summary after a list
- Code, commands, paths, error strings, numbers and negations kept verbatim
Install the ultracave skill
Download the ZIP from the skill's page on this directory. The skill is a folder that contains a file named SKILL.md. Unzip it into one of two places: your personal skills folder at ~/.claude/skills/, or a project's .claude/skills/ folder if you want the skill only for that project.
If you use the Claude apps instead, upload the skill under Settings → Capabilities → Skills.
- Personal: unzip into ~/.claude/skills/
- Single project: unzip into .claude/skills/ inside the project
- Claude apps: Settings → Capabilities → Skills, then upload
Turn the skill on and off with commands
The skill is invoked only by typing /ultracave or /caveman ultra. Natural requests such as "be brief" are not listed as triggers in the source, so use the slash forms to start the mode.
To leave the mode, say "stop caveman" or "normal mode". While it is on, you can type /caveman status to see the current mode. The skill's source says this command reports the mode and changes nothing.
- Start: /ultracave or /caveman ultra
- Stop: say "stop caveman" or "normal mode"
- Check: /caveman status
/ultracaveExamples of terse replies
The skill's own examples show the difference. A normal answer about a component re-rendering becomes a short list of causes. A question about a null check gets a one-word start followed by the next action.
Keep in mind that terse fragments can be easy to misread. The skill itself says to switch to plain prose when there is a security warning, an irreversible action to confirm, or any fragment that could be read two ways.
- Normal: "The component re-renders because an inline object prop creates a new reference each render." Terse: "Inline object prop, new ref, re-render. useMemo."
- Normal: "Yes, that should work, though you'll want a null check first." Terse: "Yes. Null check first."
What the skill needs and what it does not do
The skill requires a Claude Code or other SKILL.md-compatible agent environment. For the /caveman status reporting to work fully, the companion caveman plugin and its hooks are needed. Without the hook, the status command reports the mode you were in, or off, with a note that the host does not track it.
The skill does not run scripts and does not use the network, according to its listing. It is a text-style instruction file, so review it before installing if you want to confirm this yourself.
- Requires: Claude Code or another SKILL.md-compatible environment
- Companion caveman plugin and hooks for full /caveman status reporting
- Runs scripts: no
- Uses network: no
Tips for getting the most from it
Use the mode for long sessions where you already understand the subject and want dense, fast answers. Ask for a full explanation in plain prose when you need one, and the skill's rules allow the agent to break out for questions and for security or irreversible steps.
Remember that existing comments in files the agent edits are never deleted or shortened unless you ask, and that writing persisted outside the chat, such as code, commits or docs, follows full wording.
- Turn it off for learning or unfamiliar topics
- Turn it off for formal or customer-facing writing
- Expect explanations to be short and informal
What people use it for
- Reviewing long debugging sessions where you need quick, dense answers
- Getting short confirmations and next actions during a familiar coding task
- Reducing reading time when working through many small technical questions
Frequently asked questions
How do I install the ultracave skill?
Download the ZIP from this skill's page and unzip it into ~/.claude/skills/ for personal use, or into a project's .claude/skills/ folder. In the Claude apps, you can instead upload it under Settings → Capabilities → Skills.
How do I turn ultracave off?
Say "stop caveman" or "normal mode" in the chat. The mode stays on for the whole session until you do this.
Is the ultracave skill free?
The skill is published under the Apache-2.0 licence, as listed on this page. Check the licence file in the source repository for the full terms.
What do I need to run ultracave?
You need Claude Code or another SKILL.md-compatible agent environment. For full /caveman status reporting, the companion caveman plugin and its hooks are also needed.
Does ultracave run code or change my files?
No. It only changes the style of the agent's replies. It is listed as not running scripts and not using the network.
Is ultracave safe for explanations I need to understand?
It is best for readers who already know the subject. Terse fragments can hide reasoning and be easy to misread, so switch to normal mode when you need full explanations.
New agent skills, repos and AI tools — every week
The best new skills and open-source projects, the fastest-growing repos, and one free prompt. Five minutes to read. Unsubscribe anytime.
Our verdict
Ultracave suits developers and experienced users who want quick, dense replies during long coding sessions and can read clipped fragments without trouble. Beginners, people who need full explanations, and anyone writing formal or customer-facing text should choose normal replies or another tool. It only changes writing style and does not do work for you.
Download the skill