ats-detector

Work out which applicant tracking system a job uses from its application link, and give the formatting rules that system parses safely. Use when asked which ATS does this company use, will my CV get t

SKILL.md

GitHub

ATS Detector

An applicant tracking system reads the CV before a person does. Each one parses differently, and the cheapest way to fail is a layout the parser scrambles: two columns, a table, a header image, contact details in a text box. This skill identifies the system from the application link and checks the CV against what that system needs.

Part of the pm-cv bundle. Used by company-tailored-cv.

What This Skill Produces

  • The detected system, with the evidence (the address pattern) and a confidence level
  • A formatting checklist for that system
  • The file type to upload
  • A CV check: each rule marked pass or fail against a supplied CV, with the fix

Required Inputs

Ask for these if not provided:

  • The application link (the page where the "Apply" button leads, not the company homepage)
  • The CV, if the person wants it checked

Framework: Detection by Address

Address contains System
myworkdayjobs.com, wd1., wd3., wd5. Workday
boards.greenhouse.io, job-boards.greenhouse.io, gh_jid= Greenhouse
jobs.lever.co Lever
jobs.ashbyhq.com Ashby
smartrecruiters.com SmartRecruiters
icims.com iCIMS
taleo.net Oracle Taleo
oraclecloud.com/hcmUI Oracle Recruiting Cloud
successfactors.com, jobs.sap.com SAP SuccessFactors
bamboohr.com/careers BambooHR
workable.com Workable
teamtailor.com Teamtailor
personio.de, personio.com Personio
recruitee.com Recruitee
jobvite.com Jobvite
breezy.hr Breezy HR
moka or mokahr.com Moka (China)
zhiye.com (Beisen) Beisen (China)

If the link is a company domain with none of these, say the system is unknown and apply the strict rules below. Do not guess a vendor.

Framework: Parsing-Safe Rules (all systems)

  1. One column. No sidebars.
  2. No tables, text boxes, headers or footers for content. Contact details go in the body.
  3. Standard headings: Summary, Experience, Education, Skills (or the local equivalents).
  4. Dates in one consistent format, on the same line as each role.
  5. Real text, not images of text. No icons standing in for words.
  6. Common fonts. No ligatures or decorative characters in names and headings.
  7. File type: .docx is the safest; text-based PDF is accepted by most modern systems. Never a scanned PDF.
  8. File name: Firstname-Lastname-CV.docx.

System notes, where they differ:

  • Workday asks the candidate to re-enter or confirm parsed fields. Keep job titles, employers and dates very plain so the autofill is right.
  • Taleo and older iCIMS set-ups are the strictest parsers. Prefer .docx.
  • Greenhouse, Lever and Ashby generally keep the original file for the human reviewer, so layout quality matters more for the second reader.

These notes describe common behaviour, not a guarantee. Each employer configures its system.

Programmatic Helper

python3 skills/ats-detector/scripts/ats_detect.py "https://jobs.lever.co/acme/123"
python3 skills/ats-detector/scripts/ats_detect.py --json "https://acme.wd3.myworkdayjobs.com/en-US/careers/job/123"

It matches the address against the table above and prints the system and confidence. Standard library only. It does not fetch the page.

Output Format

ATS check: [company], [role]

1. System: [name] · Evidence: [the matching part of the address] · Confidence: high / medium / unknown 2. Rules for this application (checklist) 3. CV check | Rule | Pass / fail | Where | Fix | 4. Upload: [file type and file name]

Quality Checks

  • The evidence for the detected system is quoted from the address
  • An unknown system is reported as unknown, with the strict rules applied
  • Every failed rule has a specific fix
  • The file type recommendation is stated

Anti-Patterns

  • Guessing the vendor from the company's size or industry. Only the address is evidence.
  • Promising a CV will "pass the ATS". Systems rank and filter differently, and humans decide. Say it will parse cleanly.
  • Keyword-stuffing advice. Hidden white text and keyword blocks are detected and look worse to the human reader.
  • Treating a beautiful two-column design as fine. It may be fine for a human; it is a risk for the parser.

Example Trigger Phrases

  • "Which ATS is this job using? https://acme.wd3.myworkdayjobs.com/..."
  • "Is my CV ATS friendly?"
  • "How should I format my CV for Greenhouse?"
  • "Will this two-column CV get through Taleo?"

Run this skill in your browser: https://mohitagw15856.github.io/pm-claude-skills/skill/ats-detector.html