An applicant tracking system (ATS) turns your resume into fields: job title, employer, dates, skills. Recruiters then search and filter on those fields. If the parse goes wrong, or your resume says "Postgres" when the recruiter searches for "PostgreSQL", you can be filtered out even though you're qualified.
For 0.5.2 I rewrote the instructions resumeBuddy gives the AI so that tailored resumes are built for that first screen. One rule still sits above all the others: nothing gets added that isn't already in your own resume.
A layout parsers can't misread
Tailored resumes, and resumes built with the wizard, now follow one predictable layout. Multi-column layouts and creative headings are among the most common reasons systems like Workday and Greenhouse misread a resume, so resumeBuddy sticks to the boring version:
- Your name, then one plain-text line of contact details, with links written out instead of hidden behind clickable words.
- Standard section names: Summary, Experience, Education, Skills. No "My Toolkit".
- Every job written the same way: title and company on one line, location and dates on the next, in one date format throughout. It never adds a month your original resume didn't include.
- Simple bullets. No tables, columns, icons, or decorative symbols.
The layout also matches what resumeBuddy's PDF export can render, so the PDF you download never shows stray formatting symbols. Those symbols could also end up in the text an ATS extracts.
The posting's exact words, when they're true for you
Recruiters filter on exact terms, and most systems don't count synonyms as a match. So resumeBuddy now:
- Uses the posting's spelling for any skill or tool your resume already supports. If you wrote "Postgres" and the posting says "PostgreSQL", the tailored version says "PostgreSQL".
- Shows skills in use. Key terms appear inside an accomplishment, not only in a skills list. Newer screening tools give more weight to a skill shown with a result.
- Spells out acronyms once, like "Search Engine Optimization (SEO)", so a search for either form finds you.
- Puts your strongest matches first, in the summary and at the top of your most recent relevant job.
Cover letters do the same for the posting's two or three most important requirements, when you meet them. Some systems scan cover letters too.
What it won't do is game the filter. It won't add a keyword your resume doesn't support, even to the skills list. It won't repeat keywords unnaturally, and it won't copy sentences from the posting. The major ATS vendors now detect keyword stuffing and hidden text. And a keyword that gets you past the filter but not through the interview hasn't helped you.
Honest titles, honest years
Some systems weigh how closely your job title matches the posting's. resumeBuddy uses the posting's title in your summary only when your actual titles and scope match it. Otherwise it keeps yours.
When a posting asks for a minimum number of years, your summary states your real number, counted from the dates on your resume, instead of echoing the posting's "5+ years". If your dates don't make the number clear, it leaves the number out.
A heads-up on knockout questions
Many application forms ask automatic yes/no questions before anyone sees your resume. They cover things like a required degree, a license, minimum years, location, work authorization, or security clearance, and the wrong answer ends the application on the spot. The match check now calls these out. For each one, it says whether your resume seems to be missing it or just doesn't mention it. It also lists missing keywords word for word, so you can add the ones that are true for you.
How I tested it
Every change has automated checks, and the full extension and web app test suites pass. I also ran the new instructions against real sample resumes and job postings using a small AI model on my own machine, and kept adjusting them until the output held up. That caught two problems, both fixed before release:
- The AI added commentary like "fulfilling CI/CD experience requirements" to bullet points.
- The AI copied the posting's "5+ years" instead of stating the candidate's actual experience.
Larger hosted models generally follow instructions more closely than the small local one I tested with.
What didn't change
Your resume, your API key, and everything you generate still stay in your browser. Nothing goes through a server I run. There are no new permissions. The web app already has the update. The extension version is in review on the Chrome and Firefox add-on stores and will update automatically once it's approved.