What WROITER catches.

Looks for a fixed vocabulary list. It needs at least two distinct stronger terms, or a larger cluster containing weaker terms; a lone “landscape” stays quiet. The complete list and thresholds are in the engine source.

BANNED_WORDS · lexical

Full example — paste into the diagnostic

We delve into a multifaceted tapestry of pivotal opportunities while the team waits for a simple answer about the deadline.

One match from the engine’s phrase-template list is enough. It catches specific wording such as “it is important to note,” rather than deciding whether any sentence is generic. The contextual catalog adds specific launch, marketing, future-hype and participial templates. Ordinary individual words do not trigger these phrase rules.

BANNED_PHRASES · lexical

Full example — paste into the diagnostic

It's important to note that the files are available today and the deadline for the final report is Friday.

Flags at least two matches from a fixed set of hedge phrases, including “it could be argued” and “it seems that.” The catalog also reviews adjacent stacks such as “could potentially possibly.” Keep the uncertainty that the evidence requires.

HEDGING · lexical

Full example — paste into the diagnostic

It could be argued that the deadline needs to change. It seems that the client may prefer another week to review the draft.

Needs at least three listed intensifiers and a density above three per 1,000 words. A single “remarkably” does not trigger this detector.

EMPTY_INTENSIFIERS · lexical

Full example — paste into the diagnostic

The update is incredibly useful, remarkably effective, and undeniably significant for teams that need to export a report before Friday.

Needs at least four listed formal transitions and a density above four per 1,000 words. Matching is case-sensitive in the current engine.

TRANSITION_OVERUSE · lexical

Full example — paste into the diagnostic

However, the team needs another day. Moreover, the client requested changes. Furthermore, the editor found two errors. Therefore, we moved the release.

One listed absolute, such as “every” or “never,” must recur at least eight times and more than twice per 1,000 words. It measures repetition of a single word, not total absolute-word density.

ABSOLUTISM_OVERUSE · lexical

Full example — paste into the diagnostic

Every team reviews every draft, every editor checks every fact, every client approves every change, and every manager signs every release before publication.

Uses specific abstract combinations, clusters and repetition gates. “Load-bearing assumption” can trigger review; a literal load-bearing wall stays quiet. Weak/common vocabulary needs a cluster or repetition with a density gate. A separate low-confidence rule reviews unscoped universal claims, with negation, conditional, partitive and scoped-subject exclusions. The full candidate lists and individual thresholds are in the public specification.

CONTEXTUAL_VOCABULARY · lexical

Full example — paste into the diagnostic

This is a load-bearing assumption for the argument. The draft describes a change to the service and explains what a reader can do with it.

User-supplied post: one unscoped universal claim is enough for low-confidence review. It does not require eight repetitions.

AI visibility tip for today:

Every customer review tells AI a little more about your business.

Existing regression fixture: abstract combinations, rather than “genuine” alone.

The team claims genuine utility and meaningful engagement. The draft describes a change to the service and explains what a reader can do with it.

Existing regression fixture: three distinct listed metaphors meet the cluster gate.

Our north star is the source of truth and the backbone of the strategy. The draft describes a change to the service and explains what a reader can do with it.

Checks the first text block for a short list of announcement-style opening templates, such as “In this article” and “We will explore.”

META_INTRO · meta

Full example — paste into the diagnostic

In this article, we will explore five practical ways to plan the launch and avoid delays during the final review.

Looks in the final third of the text for conclusion markers such as “In conclusion,” “To sum up,” and “The key takeaway is.”

META_OUTRO · meta

Full example — paste into the diagnostic

The team will finish the draft on Tuesday. An editor will verify the facts on Wednesday. The client will approve the final version on Thursday. In conclusion, the report will be ready on Friday.

Flags three or more listed sequence markers, including “First,” “Second,” “Finally,” and “Furthermore.” These matches are case-sensitive.

OVER_SIGNPOST · meta

Full example — paste into the diagnostic

First, collect the draft from the writer. Second, check the facts with the editor. Third, send the final version to the client.

Matches assistant-style opening cues or a fixed set of helper-framing phrases. One cue is enough to raise this flag.

ANSWER_SCAFFOLDING · meta

Full example — paste into the diagnostic

Certainly! To help you decide, the following criteria explain which report format will keep the original headings and links intact.

The first block must start with an “Introduction” heading and contain an explicit essay or thesis announcement. Either cue alone is insufficient for this detector.

ESSAY_THESIS_ANNOUNCEMENT · meta

Full example — paste into the diagnostic

Introduction: This essay will discuss the reasons why editors need time to verify the facts before the client approves the final draft.

Needs two or more listed summary markers. The current matcher recognizes forms such as “Overall,” and “In essence,” with their punctuation.

MICRO_SUMMARY · meta

Full example — paste into the diagnostic

Overall, the team needs a clear deadline. In essence, the client should know when feedback is due before the final draft goes to review.

Matches specific reveal labels or prefaces, such as “The takeaway:” and “this matters because.” One match is enough. Catalog hooks include “The kicker?”, “Plot twist:” and “Let that sink in.” These are matched forms, not a judgment about the insight itself.

TELEGRAPHED_REVEAL · meta

Full example — paste into the diagnostic

The takeaway: the export button now downloads the entire report in one file and keeps all of the original headings intact.

Needs at least four three-item constructions and a density above three per 1,000 words. It matches short word groups in the “A, B, and C” form; a single list of three stays quiet.

RULE_OF_THREE · structure

Full example — paste into the diagnostic

Build for speed, clarity, and control. Plan for growth, change, and resilience. Deliver with care, focus, and precision. Write for readers, editors, and teams.

Needs at least two colon-introduced comma-list constructions ending in “and” or “or,” followed by a period.

COLON_LIST · structure

Full example — paste into the diagnostic

The priorities: speed, clarity, and control. The requirements: growth, change, and resilience. The team needs a clear deadline before the report is published.

Counts repeated first-two-word sentence openings. Most need five occurrences; common openings such as “It is” and “We have” need seven.

OPENER_REPETITION · structure

Full example — paste into the diagnostic

This tool checks the draft. This tool names the patterns in the copy. This tool shows the findings. This tool prepares the revision instructions. This tool leaves the final decision to the writer.

Needs four or more sentences starting with listed subordinate openers, at a density above four per 1,000 words.

SUBORDINATE_REPETITION · structure

Full example — paste into the diagnostic

While the team checks the draft, the client reviews the brief. Although the editor needs more time, the deadline remains Friday. Despite the delay, the writer finishes the revisions. Given that the facts are checked, the manager approves the release.

Needs at least six sentences, four recognized passive constructions, and a passive rate above 35%. The matcher uses listed auxiliaries and participles; it is not a full grammar parser.

PASSIVE_OVERUSE · structure

Full example — paste into the diagnostic

The draft was written on Tuesday. The facts were checked by the editor. The final version was approved by the client. The report was sent to the manager. Review took one day. Publication happens on Friday.

Needs four or more paragraphs averaging 2.5–5 sentences, with very little variation in sentence counts (relative standard deviation below 0.2). It counts sentences, not paragraph word lengths.

PARA_UNIFORMITY · structure

Full example — paste into the diagnostic

The team starts the review on Tuesday. The editor checks two claims. There is one link to fix.

The client reads the revised draft. Feedback arrives the next morning. The writer corrects the opening.

The manager confirms the publication date. A designer checks the layout. The headline needs another pass.

The final version goes to production. Everyone receives the release note. The archive keeps the earlier draft.

Needs at least two overlapping four-sentence windows whose word counts differ by no more than three, plus low overall sentence-length variation. Dialogue-heavy passages are suppressed.

UNIFORM_RHYTHM · structure

Full example — paste into the diagnostic

The team reviews the draft before each release. The editor checks the facts before each update. The client approves the copy before each launch. The writer checks the links before each handoff. The manager confirms the date before each submission.

Counts numbered and bullet markers: three numbered items, four bullets, or a mixture of at least two of each will trigger it.

OUTLINE_LIST_FORMAT · structure

Full example — paste into the diagnostic

1. Collect the draft from the writer.
2. Check the facts with the editor.
3. Send the final version to the client before the deadline.

Needs at least two list items starting with a bold label or a capitalized label followed by a colon.

LABELED_LIST_FORMAT · structure

Full example — paste into the diagnostic

- **Efficiency:** The tool helps teams finish the work before Friday.
- **Flexibility:** The tool helps teams change the plan after the editor completes the review.

Recognizes assignment-style labels at line starts, such as “Introduction:”, “Paragraph 1:”, and “Conclusion:”. One label is enough.

SECTION_LABEL_SCAFFOLDING · structure

Full example — paste into the diagnostic

Introduction: The team needs a deadline before the client can review the draft.

Conclusion: The editor should confirm the final date before publication.

Needs at least three paragraphs of 55–220 words, with the overall paragraph average in that range. Quotation/dialogue-led paragraphs and recognized citation tails suppress it. It measures block shape, not the quality of the argument.

SEGMENTED_EXPOSITORY_BLOCKS · structure

Full example — paste into the diagnostic

The first stage of a review starts when the writer sends a complete draft to the editor. The editor compares its claims with the source material and notes any questions that need an answer. A shared document lets the writer respond in place. The team agrees on a deadline so the review does not drift into the publication window.

The second stage gives the client a chance to check the draft against the brief. A client may spot a missing example or a term that the audience does not use. The writer records those changes before making another pass. A clear point of contact keeps conflicting requests from reaching the editor as separate sets of instructions during the same review.

The final stage prepares the revised draft for publication. The designer checks headings and links in the layout while the editor verifies the changes made after client review. The manager confirms that the release date still works for the team. A saved version of the draft makes it possible to compare the published copy with the approved document if a question comes up later.

Recognizes listed negative-to-positive contrast shapes. Alongside “not X, but Y,” the catalog covers split sentences, named subjects, same/different-verb reframes, “can’t / but can,” “less about / more about” and paired Stop/Start commands. Different-verb and coaching matches are low-confidence review candidates. Boundaries and length limits keep these rules narrower than any contrast in English.

PIVOT_CRUTCH · rhetoric

Full example — paste into the diagnostic

This isn't a tool. It's a revolution. The export button now downloads the entire report in one file.

User-supplied post: the split “isn’t about / it’s about” form.

AI visibility isn't about getting AI to notice you.

It's about building a reputation AI can confidently recognize✅

That's the difference.

User-supplied post: a repeated-verb reversal.

AI doesn't need more content. It needs better signals

👉Clear FAQs. Better documentation. Authentic reviews. Strong product pages.

Make your business easier to understand, and AI is more likely to understand it too.

User-supplied post: a different-verb reframe; review, rather than an automatic rewrite.

AI doesn't discover brands.

It confirms them.

Reviews. Articles. Community discussions. Industry mentions.

Each signal adds another layer of trust.

By the time someone asks AI for a recommendation, the web has already been telling your story.

User-supplied post: both commands are required for the coaching rule.

Stop asking:
"How do we rank in AI search?"

Start asking:
"Why would AI recommend us?"

The second question leads to better answers and better marketing.

User-supplied post: the “can’t / but can” form.

You can't control what AI says.

But you can control what AI sees👀

Reviews. FAQs. Helpful content. Community discussions⭐️

Better signals lead to better understanding and better chances of being recommended.

Needs two or more listed attribution phrases. It matches wording; it cannot verify whether a source exists elsewhere in the document. A catalog match such as “Independent testing confirms” can raise review on its own. Look for the actual test or source before making an edit.

WEASEL_ATTRIBUTION · rhetoric

Full example — paste into the diagnostic

Experts say this works. Studies show it matters. The report does not name a researcher or link to a source for either claim.

Needs at least two listed balance markers, such as “on one hand” and “on the other hand.” It cannot determine whether the opposing positions are equally credible.

FALSE_BALANCE · rhetoric

Full example — paste into the diagnostic

On one hand, the team wants to publish the report on Friday. On the other hand, the editor needs another day to verify the facts.

One listed discovery or concealment phrase is enough, including “buried in the documentation” and “quietly slipped.” It cannot establish whether something was actually concealed. The catalog also reviews a lagging-crowd claim, unsupported novelty framing, and the combined crowd + correction + lesson closer. The “smug mentor” rule requires all three cues within bounded text; a single teaching sentence is insufficient.

MANUFACTURED_DRAMA · rhetoric

Full example — paste into the diagnostic

Buried in the documentation, the update adds an export button that downloads the complete report in a single file for the editor.

User-supplied post: the combined crowd, correction and lesson frame.

Many businesses focus only on AI recommendations.

But first, AI has to:

✓ Recognize your brand.
✓ Understand your expertise.
✓ Trust you enough to recommend it.

That's how AI visibility works.

Matches a narrow run of at least three short “No…” sentences. It also recognizes a specific clipped-praise trio and at least two repeated “Not always. Not perfectly.” pairs. It does not flag every short sentence or every social post.

STACCATO_DRAMA · rhetoric

Full example — paste into the diagnostic

No fuss. No friction. No limits. The tool handles the queue. The draft describes a change to the service and explains what a reader can do with it.

One match from the assistant-voice phrase list raises the flag. It recognizes particular greetings, offers of help, and sign-offs. The catalog also recognizes generic affirmation such as “You’re absolutely right” and subjectless helper cues such as “Happy to help.” Curly apostrophes are normalized.

ASSISTANT_PERSONA · persona

Full example — paste into the diagnostic

Great question! I would be happy to explain how the export button downloads the entire report in one file for your editor.

One recognized self-disclosure or training-data phrase raises the flag, including “as an AI” and “my knowledge cutoff.”

AI_DISCLAIMER · persona

Full example — paste into the diagnostic

As an AI language model, I cannot access real-time information about the report, but the editor can check the source before publication.

One listed credibility-preface match raises the flag. It detects the wording, not whether the writer is honest. Sentence/line-opening “Honestly,” and “Quite frankly,” and named candor labels such as “My honest take” receive review. “An honest person” and “She answered honestly” stay quiet for these rules.

CREDIBILITY_THEATER · persona

Full example — paste into the diagnostic

Let me be direct: the export button now downloads the entire report in one file and keeps all of the original headings intact.

Existing regression fixture: a candor preface, not a ban on “honestly.”

Honestly, the queue needs another worker. The draft describes a change to the service and explains what a reader can do with it.

Existing regression fixture: a named candor label. An ordinary caveat is contextual.

The honest caveat: the queue can stall. The draft describes a change to the service and explains what a reader can do with it.

Needs two distinct self-praise cues, or one cue accompanied by at least four “we/our” sentence openings and a strong skew toward the maker. First-person writing alone is insufficient.

SELF_CONGRATULATION · persona

Full example — paste into the diagnostic

Our transparency principle guides the work. Unlike most tools, we offer an honest account of the process because users deserve to know how the report was prepared.

Recognizes listed citation/container tokens, unfilled placeholders and named AI-referrer URL parameters. Code fences, inline code and ordinary link destinations are protected; AI-referrer checks inspect URLs specifically.

CHATBOT_RESIDUE · persona

Full example — paste into the diagnostic

The queue can stall. contentReference[oaicite:0]{index=0} The draft describes a change to the service and explains what a reader can do with it.

Needs at least three matches from a fixed set of replacements such as “serves as,” “stands as,” and “acts as.” A single occurrence stays quiet.

COPULA_AVOIDANCE · style

Full example — paste into the diagnostic

The report serves as a guide. The guide stands as a resource. The dashboard acts as a hub for new users.

Reviews repeated em dashes, bold spans or hashtags when both a minimum count and a density gate are met. These are formatting-specific rules; smart quotes and a lone dash stay quiet. Individual thresholds are listed in the public specification.

FORMATTING_OVERUSE · style

Full example — paste into the diagnostic

The **queue** takes **files** from **clients** and sends **jobs** to **workers** before **dinner**. The draft describes a change to the service and explains what a reader can do with it.