Score your AI coding agent setup

Everyone else counts what you spent. This counts what you built — skills, hooks, subagents, MCP servers, the rules you wrote. Level 1 to 80.

npx vibecoder-skills

Runs offline. Sends nothing. Prints your level in about 150 ms.

Seven axes, and spending is the smallest

Token spend is worth at most ten points out of a hundred, on a log scale. Every axis has its own ceiling — a thousand empty skills score the same as forty real ones. Level 80 needs an essentially flawless setup and is supposed to be rare.

Toolkit
24
Automation
18
Discipline
14
Safety
14
Scale
12
Depth
8
Burn
10

Your setup never leaves your machine

  • No paths, no project names, no repository names — only how many.
  • No prompts, no transcripts, no code.
  • Private skill names are counted, never listed.
  • Your id is a random UUID, not a hardware fingerprint. No IP is stored.
  • Nothing is scheduled and nothing runs after the command exits.

Enforced by a test, not by this paragraph — and a second test breaks the redactor on purpose to prove the first one would catch a leak.

Top setups

#1
typex
TH
Sovereign
63