Skip to content
Zevqi

Free

Free AI code risk audit

One command shows how much of your codebase rests on too few people. It reads git history on your machine and sends nothing.

Run it

From inside any git repository

npx zevqi audit --html

It prints a read-out and writes zevqi-audit.html: one self-contained page with no scripts and no outside requests. Open it, send it to a colleague, or print it to PDF. Add --months 24 to look further back.

What you get

Four things a leader can act on

Files one person holds

How many source files have only ever been changed by one person, and how many of those sit in authentication, billing, infrastructure or data code.

Owners who went quiet

Files whose main author has not committed for 90 days or more: the ones that would be hardest to change if something broke.

Where knowledge is thinnest

The folders with the most files resting on one person, so you know where to put a second pair of eyes first.

The largest single dependency

How many files, and what share of the code, rest on the one person the most depends on. A count, not a name, unless you ask for the name.

Honest limits

What it cannot tell you

  • It does not read your source code, send anything over the network, or need an account, a team or a model key.
  • It does not say who understands a file. Git shows who changed it. Whether the people who are left can explain it needs review evidence, which the team view records.
  • It does not score or rank people, and it is not a compliance certificate.

AI-assisted work is counted only where a commit says so, so those figures are a minimum. The rules are printed in the report, and every flag says why it was raised.

Want to know who actually understands it?

The team view records review evidence on each AI change, shows which files are covered and which depend on one reviewer, and answers what would be left if someone went.