Review TrueProfit Live Blog URLs
Review the published article itself. Do not compare it with a Google Doc unless the user explicitly requests a comparison.
Scope
- Start at the article H1.
- End immediately before the author bio identified by
div.wrap-bio. - Include the FAQ component even when it is streamed into a placeholder immediately before the author bio.
- Exclude navigation, byline/profile chrome, newsletter forms, sidebar elements, author bio, Related Blogs, footer, and other site chrome.
- Review only the live URL or URLs provided by the user.
Required extraction method
TrueProfit pages use streamed React components. Markdown extraction commonly omits Quick
Recap, Further Reading, FAQs, notes, examples, and highlighted boxes. Raw HTML may place the
resolved component payload after the article even though a template placeholder locates it
inside the article.
- Fetch the page as readable markdown to inspect the main article.
- Fetch and inspect the raw HTML before concluding that any custom component is absent.
- Reconstruct streamed placeholders before applying the article boundary:
- A placeholder such as
template id="B:0"marks the component's true location. - Its resolved payload is normally under a later element such as
div id="S:0". - Treat the contents of
S:0as if they appear atB:0; repeat for every matching number.
- A placeholder such as
- Only after reconstruction, evaluate the content from H1 through the pre-author-bio boundary.
Prefer scripts/extract_live_article.py <URL> for this extraction. If the script cannot run,
perform the same raw-HTML verification manually. Never infer absence from markdown alone.
Cross-check the extracted arrays against the reconstructed article. FAQ headings may use
Frequently Asked Question(s) instead of FAQs, and questions may lack a special CSS class.
An empty array is not proof of absence when the article still contains the relevant component.
Inspect its questions and answers manually if necessary. Confirm that the author-bio boundary
was found before reporting a complete audit.
Default review
Always report all five sections below. The absence of an element is a valid result, but only after raw-HTML verification.
1. Heading structure
Check:
- exactly one article H1;
- H1-H6 hierarchy and skipped or orphaned levels;
- complete and correctly ordered numbered headings;
- headings used at a reasonable semantic depth;
- materially repetitive headings that target the same intent.
Suggest replacements only for unreasonable or repetitive headings. Do not provide general grammar or copy edits.
Check semantic nesting, not only heading levels: an H3 must belong under its parent H2. For example, report-navigation instructions do not belong under a formula-only H2; recommend an appropriate H2 or a move to a relevant section. Check every independently numbered group. A short overview followed by a detailed section is not automatically redundant.
Flag conflicting years between the H1 and other headings, including FAQ headings, as a heading consistency issue. Do not treat a suggested year alignment as factual verification or assume that changing the year alone updates the answer. Ignore extraction-only spacing artifacts unless they are confirmed in the rendered heading.
2. Quick Recap
Report whether it exists, its number of bullets, and its location relative to the introduction and first H2. Do not judge claims or rewrite the recap unless the user asks for content review. An explicit request for a replacement recap also authorizes that narrow rewrite; base it on the current article rather than a generic template.
3. Further Reading
List every Further Reading box with its linked titles and its position between surrounding headings.
Use a Markdown table with Box, Links, Placement, and Assessment columns.
Count H2s in the current live article, not a proposed revised outline. Verify whether ordinary
body text separates a box from the next heading; the extractor's nearest-heading fields alone
do not establish that the box is directly above that heading.
Warn only when a box sits directly above the 2nd, 3rd, 4th, or 5th article H2, excluding a Quick Recap heading if one exists. Do not warn when it is above the first or 6th-and-later H2, above an H3, or separated from the next H2 by ordinary body text.
Do not audit missing or duplicate internal links.
4. FAQ redundancy
List every FAQ question and classify it as:
Heading duplicate— same main intent as an article heading;Answered in body— the body already provides the answer;Partial— related to the body but adds a meaningful sub-angle;Clear— genuinely new and useful.
For redundant questions, recommend removal or a concrete uncovered re-angle. Prefer a re-angle when the topic is useful. Do not flag mere keyword overlap.
Compare each complete FAQ answer with the body, not just its question. A platform-specific or
technical sub-angle may be Partial or Clear even if it shares keywords with a main heading.
Do not recommend an alternative already covered elsewhere. A Clear classification or
Keep recommendation concerns redundancy only; it does not endorse factual or legal accuracy.
Present the FAQ review as a Markdown table with exactly these columns:
FAQ, Classification, and Recommendation. Put each FAQ question in its own row; do not
use a numbered list for this section.
If there is no FAQ component after raw verification, say so plainly.
5. Image alt text
Within the article scope, flag only images whose alt attribute is missing or empty.
- Do not evaluate whether existing alt text is well written.
- Do not flag automated images with
alt="banner cta". - Do not flag lazy-loading placeholders with
alt="Loading...". - Do not audit CTA eligibility, CTA placement, or tracking links unless explicitly requested.
Identify each missing-alt image by its section or nearby heading. When asked to create alt
text, inspect the actual image instead of inferring its content from a filename or nearby copy.
Give the suggested alt text as plain, copyable lines in article order, without alt="..."
wrappers, quotation marks, numbering, or extra commentary when the user asks for alt text only.
If an image cannot be inspected, say so rather than inventing a description.
Optional content and factual review
Run this pass only when the user explicitly asks for content review, factual review, accuracy,
fact-checking, sources, references, benchmarks, or current information. Otherwise, do not
include a Content and factual issues section and do not browse for claim verification.
If the user has not said whether they want this pass, complete the five default review sections
and end by asking: Would you also like me to check content and factual issues for this article?
Do not run the pass until the user confirms. If the user already requested it, include it in the
same review. If the user explicitly declined it, omit it and do not ask again for that article.
For a continuing series of "Next" URL reviews, carry forward the user's established preference to exclude content and factual issues until they change it; do not repeat the confirmation question for every URL. Do not infer authorization for fact-checking from "next", "review", or a request for structural QA. A request limited to one article or one rewrite does not automatically enable factual review for subsequent articles.
When requested, read and follow references/content-factual-review.md.
Excluded by default
- Google Doc comparison
- missing or duplicate internal-link opportunities
- general copy, grammar, punctuation, or style corrections
- external-reference or benchmark verification
- CTA-image eligibility and CTA tracking-link review
- a generic
Priority findingssection
Response format
Lead with the live URL reviewed and whether the optional content/factual pass was included. Use these headings in this order:
Heading structureQuick RecapFurther ReadingFAQ redundancyImage alt text
Add Content and factual issues only when explicitly requested. Keep the report concise,
actionable, and advisory; do not edit the live site. When the user's preference for the optional
pass is unknown, finish with the confirmation question defined above.