Skip to content

What each pass is looking for

Five passes, five different jobs. Learn to route a defect to the right one and you never run the wrong check twice.

9 min read

Level 0

Curious

0 XP
150 XP to ApprenticeNo streak yet0 badges0% done

The passes are not five reads of the same thing. Each one is a different question, asked in a different mode.

PassThe questionLooks for
1. Content and slopDoes this sound like a person?Kill-list words, staccato, contrastive negation, claims with no number, walls of text
2. MechanicsDoes it follow the format rules?Em dashes, curly quotes, spelling variant, heading case, anchor length, prose sandwiching, list caps
3. Fresh readDoes the argument hold?Contradictions, a promise in the intro the article never keeps, orphan pronouns, sections that go nowhere
4. ClusterWhat do the siblings say?Repeated phrasing across the batch, a source cited three times, a missing CTA or link set
5. Fact and ruleIs any of this not allowed?Unverified numbers, claims the knowledge base rules out, client policy violations

Run one pass at a time

Name the pass you are running

Asking for "a proofread" gets you a blur of all five done shallowly. Asking for pass 2 only gets you an actual mechanics check, because the model is not also trying to judge the argument.

claude
>Run pass 2 on the draft. Mechanics only. Do not touch the prose.
Skillproofread-articlepass 2 of 5
Bash(python references/article-check.py draft.md)
em dashes: 1 (line 214) curly quotes: 0 UK spellings: 2 ('optimise' x2) section ends on rich format: 1 ('Cycle counting') body lists: 4 (cap is 3) anchors over 5 words: 2
Six findings in under a second, and all six are things a human reader skims straight past.
Six mechanical issues. Fixing each, then re-running the check.
Bash(python references/article-check.py draft.md)
All checks clean.
Challenge

Route the defects

130 XP+33 perfect

Six real findings from one batch. Send each to the pass that would actually catch it. Route all six correctly and you unlock the Five passes badge.

1Content & slop2Mechanics3Fresh read4Cluster5Fact & rule

Section 3 contains "leverage their data to enhance operational efficiency."

The article uses "optimise" twice. Every other article on the site uses "optimise" nowhere and "optimize" throughout.

Paragraph two says forecasting accuracy matters more than speed. Paragraph nine concludes that speed is what actually moves the number.

Three of the four articles in this batch cite the same 2026 industry study for their opening statistic.

The article recommends the product for warehouse management. The knowledge base says warehouse management is explicitly out of scope.

A section ends on a five-row table with no closing sentence, and the next heading follows immediately.

Finished reading?

Mark it done to bank the XP and keep your streak alive. Any challenges on this page score separately, so you can come back for a perfect run later.