What ATSPrep measures—and what it does not.

PDF files are opened with PDF.js and converted into positioned text items. DOCX files are read from their document XML. Plain-text and Markdown files are read directly. ATSPrep then maps recognizable headings, contact fields, roles, dates, education, and skills into editable resume fields.

Import is best-effort. Unusual headings, graphics, scanned images, complex tables, and multi-column layouts can produce incomplete or incorrectly grouped results.

Enterprise parsing vendor Textkernel similarly distinguishes text-based PDFs from scanned files that require OCR and recommends sending an original document rather than converted or copied text. See its supported file-format guidance.

The Proof Engine applies deterministic rules to each bullet. It looks for measurable evidence, weak openers, filler phrases, vague quantifiers, passive voice, first-person language, and unusually short or long bullets.

• It does not use a language model or invent achievements.
• A higher score is not a guarantee that a recruiter will prefer the writing.
• Quantification is not appropriate for every role; users should keep accurate, defensible claims rather than manufacture numbers.

ATSPrep generates a single-column PDF with standard fonts, real text, conventional headings, and top-to-bottom drawing order. It avoids images, text boxes, and tables for core resume content.

This follows the broader accessibility principle that meaningful content should remain available as text and in a logical reading order. See the W3C PDF accessibility techniques.

The PDF Association explains that Tagged PDF can define semantic structure and logical ordering independently from visual layout. ATSPrep does not claim to produce a fully tagged accessible PDF, but it follows the same principle of preserving ordered text. See Techniques for Accessible PDF.

After generating the PDF, ATSPrep opens those exact bytes with PDF.js, reconstructs visual lines from positioned text, and compares the recovered output with the resume source.

• Checks whether a usable text layer exists.
• Checks name, email, phone, headings, bullets, ordering, encoding, and page count.
• Reports the share of normalized source tokens found in the extracted text.

PDF.js is an open-source PDF parser maintained by Mozilla. Its official project page explains its role in parsing and rendering PDF documents.

ATSPrep does not connect to or emulate the proprietary ranking, parsing, or workflow configuration of Workday, Greenhouse, Lever, Taleo, iCIMS, Ashby, SmartRecruiters, or any employer. Those systems can differ by version and customer configuration.

A successful X-Ray means the generated PDF contains recoverable text in a sensible order when read by PDF.js. It does not prove keyword relevance, recruiter preference, employer eligibility rules, or compatibility with every ATS.

• Mozilla PDF.js — the parser used for local PDF extraction.
• W3C PDF Techniques for WCAG 2.2 — text, structure, and reading-order accessibility techniques.
• PDF Association: Techniques for Accessible PDF — how logical structure and ordering support extraction and reuse.
• Textkernel supported file formats — parser-vendor guidance on text PDFs, scanned files, and OCR.
• Adobe: Create and verify PDF accessibility — guidance on searchable text, document structure, and reading order.
• MDN Web Docs: Using files from web applications — browser-side file access used by local import workflows.

If a description is inaccurate or a result is reproducibly wrong, email support@atsprep.com. Methodology changes will be reflected on this page.