resume-export-standards
ATS-safe + accessible resume/CV export standards (PDF/DOCX) — real ATS parsing limits, single-column rules, document-metadata accessibility (full PDF/UA-1 tagging is a future goal), country/industry CV norms (US/UK/DE), keyword hygiene. Load for changes under export/, model/, theme/, templates/, locale/, fonts, layout/.
saeedkolivand · 已有 AI 审核记录
安装与使用
运行命令前先阅读来源与依赖说明。复制命令不会自动安装或执行。
npx skills add saeedkolivand/ai-job-hunter-app --skill resume-export-standards技能说明
name: resume-export-standards description: ATS-safe + accessible resume/CV export standards (PDF/DOCX) — real ATS parsing limits, single-column rules, document-metadata accessibility (full PDF/UA-1 tagging is a future goal), country/industry CV norms (US/UK/DE), keyword hygiene. Load for changes under export/, model/, theme/, templates/, locale/, fonts, layout/.
Resume/CV export standards — ATS-safe & accessible
External best-practices for generating resumes that real ATS parse and that meet accessibility law. Load with author-contract (pdf-docx-generator) / token-efficiency (resume-export-expert). Pairs with the repo's docs/knowledge/resume-domain.md (the export contract).
ATS-safe formatting (verified 2026-06)
Parsers read linearly, top→bottom, left→right — design for that.
- Single-column, reverse-chronological. Multi-column / side-by-side scrambles reading order into "word salad." https://www.jobscan.co/blog/resume-tables-columns-ats/
- No tables/text-boxes/grids for layout. Modern ATS tolerate a simple one-cell table but tab stops are safer; skills as commas/bullets/
|, never a grid. - Contact info in the BODY, never headers/footers — ATS dropped header/footer contact info ~25% of the time. https://www.jobscan.co/blog/ats-formatting-mistakes/
- Standard fonts (Arial, Calibri, Helvetica, Times New Roman, Georgia, Garamond, Cambria, Verdana), 10–12pt; custom display fonts → fallback-render breakage.
- Standard
•/-bullets only — icon glyphs/dingbats parse as[NULL]; replace icons with text labels ("Phone:" not 📞). - No graphics/charts/logos/photos/skill-bars (US/UK/CA — see country norms); images are invisible to parsers.
- Standard section headings ("Work Experience", "Education", "Skills") — creative titles ("My Journey") aren't recognized.
- PDF vs DOCX: text-selectable PDF now parses ≈ DOCX in modern systems (Workday/Greenhouse/Lever); DOCX safer for older/stricter ATS; default PDF unless the posting asks for Word; never image-only/scanned PDF. https://www.resumemate.io/blog/pdf-vs-word-for-resume-2026-which-format-ats-actually-prefers/
- File name
FirstName-LastName-Role.pdf(descriptive, no spaces/special chars). - Dates consistent
Mon YYYY – Mon YYYY; no apostrophes (Jan '21), no year-only when months exist. - Keywords mirror JD terminology inside achievement bullets with metrics — don't repeat; stuffing drops scores ~30% (parsers detect repetition + understand synonyms). https://blog.theinterviewguys.com/ats-resume-optimization/
Document accessibility — current posture + PDF/UA-1 as a future goal
What we ship today: single-column, selectable-text PDFs with logical top→bottom
reading order (the same structure ATS parse cleanly), and each PDF sets document Title +
author + language metadata (#set document(title/author) + #set text(lang) in the Typst
engine) plus embedded fonts — a real screen-reader aid. We do not yet emit a full
PDF/UA-1 tag tree. Do not claim tagged/PDF-UA output that does not exist.
Future goal (not a current guarantee): full PDF/UA-1 (ISO 14289 · WCAG 2.2 AA) is an all-or-nothing standard — tag tree, per-template alt text, nested heading structure, veraPDF validation — on a still-maturing Typst feature. Park it behind its own ADR if/when committed.
The external standard, kept as the direction to grow toward:
- Tagged structure, logical reading order, real heading hierarchy (not visual sizing). https://accessibility.build/guides/pdf-accessibility
- Declare document language + Title metadata (done today); embed fonts with ToUnicode (done); bookmarks for 20+ pages (future).
- Alt text on meaningful images; decorative → artifact; data tables get header-cell markup (future, with tagging).
- Accessible DOCX (built-in Heading styles + real lists/tables + alt text) maps to PDF tags on export. https://daisy.org/guidance/info-help/guidance-training/content-creation/accessible-pdf/
- 2026 driver: EU European Accessibility Act applied 2025-06-28, enforcement ramping through 2026 (EN 301 549 → WCAG 2.1 AA + PDF/UA). https://www.levelaccess.com/compliance-overview/european-accessibility-act-eaa/
Country / industry CV norms (2026)
| Element | US / UK / CA / AU | Germany / EU |
|---|---|---|
| Photo | Omit | Expected (formal headshot, top-right) |
| Date of birth | Omit (anti-discrimination) | Often included |
| Address | City/country only | Full address standard |
| Length | 1–2 pages | 1–2 pages (Lebenslauf), often signed/dated |
- German AGG means a photo/DOB cannot be legally required, yet most DE employers still expect a photo. https://www.topcv.io/blog/cv-for-germany-complete-guide-2026
- Creative/design roles tolerate portfolios + visuals; finance/legal/gov/academic and any ATS-screened pipeline demand the plain single-column rules.
Common ATS-failure mistakes (quick reject list)
- Multi-column/table layout → scrambled reading order. 2. Contact info in header/footer → ~25% dropped. 3. Icons/photos/skill-bars/charts → invisible/
[NULL]. 4. Custom display fonts → fallback breakage. 5. Image-only/scanned PDF → zero text. 6. Creative section headings → section unrecognized. 7. Inconsistent/apostrophe/year-only dates → misparsed tenure. 8. Keyword stuffing → ~30% penalty.
Take: ATS-safe and accessible are the same single-column, text-first artifact — they reinforce, not conflict (full PDF/UA-1 tagging is the future accessibility goal, not something we emit today). (Vendors don't publish parser specs, so "simple tables parse OK" is the weakest claim — hedge it.)