expressjs/express/examples/auth: 21% comprehension debt

21% of the code we scored in expressjs/express/examples/auth — weighted by bytes, covering 100% of its code — sits in files no human has recently written, reviewed, or explained (comprehension score below 0.30 out of 1.00). Higher is worse. Non-code files — images, lockfiles, minified bundles, prose docs, environment files, generated files — are excluded throughout.

See your own repo's number — install free →free · read-only · installs in 2 minutes

Every score decomposes — hover any cell.Methodology →github.com/expressjs/express ↗

expressjs/express · comprehension mapas of 2026-07-283 of 4 files scored100% of code scored, by bytesscorer v4

Hide file names

No human on its recordRecent human contactHatched = consciously accepted riskCell size = file size · hover any cell for its factor breakdown

The paydown path

Substantively review the 2 largest indebted files — 21% of the scored code — and this repo's comprehension debt falls from 21% to 0%.

Recomputed through the same scorer, not estimated: those files are given full review depth and every other file keeps its stored score. The most impactful of them are listed under Riskiest paths below.

Why this repo scores what it does

The four factors behind every score. The gap is what this repo is leaving on the table — the biggest gap is the biggest lever. Question answerability is 0 on every file everywhere: it was answered by the people it graded, so it is recorded but not scored.

Human review depthearning 0.09 of 0.40 · gap 0.31

21% of scored code, by bytes, gets zero from this factor.

Human author recencyearning 0.04 of 0.25 · gap 0.21

21% of scored code, by bytes, gets zero from this factor.

Bus factorearning 0.22 of 0.25 · gap 0.03

0% of scored code, by bytes, gets zero from this factor.

Question answerabilityearning 0.00 of 0.10 · not scored

100% of scored code, by bytes, gets zero from this factor.

Where the debt concentrates

Each percentage is that directory's own debt — its indebted bytes over its scored bytes — so these do not add up to the repo number.

examples/auth/21% debt

Riskiest paths

examples/auth/views/login.ejs0.08/1humanAuthentication code · never substantively reviewed
examples/auth/views/head.ejs0.08/1humanAuthentication code · never substantively reviewed

Who wrote this code

Authorship is not the story. Across the gallery, the most indebted repo is almost entirely human-written and the healthiest is AI-era. Debt tracks how deeply humans review — not who, or what, wrote the code.

Human-authored100%

Share of scored bytes — the same denominator as the debt percent. Detection method is public and disputable, entry by entry.

How this is scored

Comprehension score = 0.40 human review depth + 0.25 author recency + 0.25 bus factor + 0.10 question answerability. Bus factor caps at 3 distinct humans; recency decays over 180 days. A file counts as debt below 0.30.

Question answerability is 0 for every file Fathohm scores, here and in every installed repo. It was answered by the same people it graded, so it is collected and shown back to the team but never scored — the weight is held open for a check a teammate can verify. Every score therefore tops out at 0.90, not 1.00.