Scored by the 15 published checks below — the same rubric, the same weights, every time. Same config in, same score out.
How the score works
Every config starts at 100. Each check that fails subtracts its published weight (below). The score never goes below 0. A check that doesn’t apply to your config is marked not applicable and subtracts nothing — a remote-only config isn’t punished for having no local files to scope, and a local config isn’t punished for having no remote URLs.
The battery is deterministic: it reads only the config text you paste, applies the rules printed on this page, and produces the same score for the same config every time. No calls home, no machine learning, no vibes.
What this is — and isn’t. This is a score from 15 named checks over one config file. It reads the config, not the servers’ code, and it can’t see how you actually run things day to day. A high score means these 15 openings weren’t found in what you pasted — it is not a promise that a setup can’t be misused, and a low score is a work list, not a verdict on you. Grade is deliberately conservative: when a rule can’t tell whether something is scoped or safe, it says so instead of guessing in your favor.
The rubric — all 15 checks, published
Every check, its weight, and the exact rule it applies. If a knowledgeable builder finds a rule here that’s wrong, we fix the rubric in public — that’s the deal.
| Check | Weight | The rule |
|---|
Run the whole battery, any time, on anything
This page runs 15 checks. The full J-Dawgs auditor catalog behind it — every auditor in this battery, and 300+ more — is unlocked forever with the Unlimited Pass: every priced tool in the AI Tool Store, including future ones, for one payment.
$99.99 — one time. Not a subscription.
Get the Unlimited Pass