Skills Explorer Add Skills

caveman-setup

An agent skill by JuliusBrussee, from JuliusBrussee/caveman. Tags: cloud, integration, observability, setup.

What it does

Wire a repository through the Caveman Cloud gateway so every LLM request is measured, with no behavior change. Use for "set up caveman" or adding LLM spend observability.

Install

With the skills CLI, which installs into Claude Code, Codex, Cursor and other agents:

npx skills add JuliusBrussee/caveman --skill caveman-setup

Or copy the skill folder into Claude Code's skills directory by hand (~/.claude/skills for every project, or .claude/skills inside one):

git clone --depth 1 https://github.com/JuliusBrussee/caveman
cp -r caveman/skills/caveman-setup ~/.claude/skills/caveman-setup

Safety box score

Grade D, elevated risk: 54 of 100 risk points, rated 2026-10-15. The grade rates what the skill and its scripts can reach on the machine of whoever installs it, across eight categories from shell execution to secrets access. How the score works.

This skill rewires every live LLM callsite in a repo through the Caveman gateway and then sends one real verification request. The main risks are broad network exposure of prompts/responses, handling API keys in env files/headers, and an explicit instruction to run the verification without asking for confirmation.

Category levels, from 0 (none) to 3 (high)
CategoryLevelWhy
Code & shell execution1 (low)It directs the agent to run shell commands such as `curl -sS "$CAVE_GATEWAY_URL/..."` for verification, but it does not ship native executables or install scripts.
Network access3 (high)It routes all live LLM traffic through a remote gateway (`GATEWAY/w/<app>/...`) and requires a real verification request plus optional remote docs fetching.
Filesystem reach2 (moderate)It edits repo files for configuration (`.env`, `.env.local`, and possibly `.gitignore`) and touches source callsites, but it is not asking to modify system paths.
Secrets & credentials2 (moderate)It requires `CAVE_API_KEY` in an env file and discusses provider keys in headers/env vars; it explicitly warns not to commit or print them.
Privilege & persistence0 (none)There is no sudo, service registration, global install, or OS/shell startup modification.
Instruction hijack surface3 (high)It explicitly says, `Send it now — do not pause to ask permission for it`, which pressures the agent to skip confirmation for a real external request.
Irreversible actions2 (moderate)It autonomously performs a real billable verification request and rewires all live LLM callsites, but it does not deploy, publish, or delete external resources.
Transparency0 (none)The skill file is plain text and does not rely on binaries, obfuscated blobs, or unpinned installers.

Before installing

Rated 2026-10-15 by an automated review of 1 file from the skill's repository.

Source

Repository
JuliusBrussee/caveman (all skills from this repository)
Path
skills/caveman-setup/SKILL.md
Branch
main
Updated
2026-10-05