A thin content checker helps identify pages that deserve an editorial review. Screpy measures text available in source HTML and compares repeated paragraphs across up to ten submitted pages. It flags low text volume and extensive shared wording without claiming to judge usefulness, expertise or search rankings.
Compare pages with a clear purpose
Paste one public HTTP or HTTPS URL per line and choose the content language. Duplicate normalized addresses are checked once. Use a consistent language for the batch so word segmentation is appropriate. The optional page type records context; it does not change the fixed review threshold. For a broader technical inspection of one page, use On-Page SEO Checker.
Understand what the word counts include
The tool prefers a main region, then article, and otherwise uses body-based extraction. It reports that choice because extraction from a body can include unrelated copy. Main-content words, body words and main paragraphs answer different questions. JavaScript is not executed, so source HTML may omit content that visitors later see.
Fewer than 200 main-content words prompts review; it is not a minimum imposed by Google. A short product specification or useful contact page may already serve its purpose. Google explicitly says it has no preferred word count in its people-first content guidance. Use Keyword Density Checker for a separate examination of recurring terms.
Review shared wording within the submitted list
Paragraphs with at least 15 word tokens are compared as normalized, case-insensitive word sequences. The percentage counts words in paragraphs also found on another submitted page. A value of 50% or more prompts review, not a duplication penalty. This is not semantic similarity, web-wide plagiarism detection or proof that repeated text is inappropriate.
Percentages require at least two fully analyzed pages; otherwise they remain unknown. Up to 20 observed shared-paragraph examples are retained, each limited to 300 characters. Inspect the full pages before removing legal, product or service wording that may intentionally repeat.
Keep incomplete results separate from short pages
Each HTML document is limited to 2 MiB and a six-second fetch. The batch has an 8 MiB total download limit and a 30-second API deadline. Text extraction is bounded to 200 KiB and 2,000 main paragraphs. Truncated counts stay unknown rather than presenting partial text as a complete measurement.
Filter rows and export CSV, or download the bounded JSON report. An unavailable page is not an empty or low-quality page. Investigate its HTTP response with Bulk HTTP Status Checker before deciding what content to change. The report does not check Google's index or recommend automatic deletion or noindex.