A Chrome extension that turns a job description into a tailored CV

The hard part of applying to a lot of jobs isn’t the writing. It’s the tab-juggling: selecting a description off a page that fights you, pasting it somewhere, switching to your CV, editing, exporting, naming the file. Two weeks later you have CV_final_v3.pdf, CV_final_v3_backend.pdf and CV-2026-b.pdf, and no memory of which posting any of them was cut for. When a recruiter rings, you can’t tell which version they’re holding.

SiviGen’s Chrome extension removes that and nothing else. It reads the job description off whatever page you’re on, shows it to you to trim, and sends it — the server parses the title, company and requirements, and tailors your CV to that role. It is a capture tool, not an auto-applier: it does not submit applications on your behalf.

None of the tab-juggling is the actual work. The actual work is deciding which of your real experience matters for this role and saying it well. The rest is clerical overhead that scales linearly with how many jobs you apply for — the wrong shape for the one part of job hunting where volume genuinely helps.

How to judge any job-capture extension

Most of what follows is about ours, so here is the part that isn’t. There are a lot of extensions that grab a job posting and do something with it, and four questions separate the ones worth installing from the ones worth uninstalling. None of them require you to trust a vendor’s description of itself.

Does it extract with a general algorithm or with per-site selectors? A selector-based extractor is a list of CSS paths — “the description lives in div.job-details on this board” — maintained per site. It works well on the three or four boards the developer tested and produces nothing, or garbage, on a small company’s own careers page. It also breaks the week a board ships a redesign, and you find out by getting a CV cut for the wrong text. A general-purpose article extractor like Mozilla Readability — the open-source library behind Firefox’s Reader View — scores blocks of the document by text density and link ratio instead, so it treats every site the same because it never learned the difference. It is less precise on a page it could have had a rule for, and far more robust everywhere else. For job hunting, where the interesting roles are often on obscure ATS subdomains, that trade is the right way round.

Does it show you the text before it sends it? Extraction is imperfect on any algorithm. If the tool doesn’t put the captured text in front of you in an editable box, you cannot tell whether it sent the description or the cookie banner, and every downstream output inherits the mistake silently.

When does it read the page — always, or when you ask? A content script registered on every site is normal and often necessary; a content script that transmits on every page you visit is not. You do not have to take this on faith. In chrome://extensions open the extension’s Details and look at Site access — Chrome lets you set most extensions to run only on click, which is worth doing for any of them. And you can watch the traffic yourself: right-click the extension icon, choose Inspect popup, and open the Network tab before you press anything.

Does it interpret, or just capture? An extension that decides what counts as a “requirement” inside your browser is shipping judgement that only updates when users update the extension. One that captures text and lets a server interpret it is easier to fix and easier to audit — and it means the interesting logic is somewhere you can be told about it changing.

Apply those four to this extension too. The rest of this page is the answers.

How does the SiviGen Chrome extension work?

The extension has three moving pieces and deliberately no intelligence of its own. A content script runs on the page you’re viewing and extracts the job text using Mozilla Readability, which strips navigation, adverts and related-jobs rails and keeps the article body. A popup shows you that text in an editable box, so you can trim a stray footer or paste something in by hand before anything is sent. When you press send, a background service worker posts the raw text to SiviGen, and the server does all the structuring: it parses the job title, the company and the individual requirements, then runs the tailoring pipeline against your CV. The extension never decides what a “requirement” is. It captures text and hands it over.

That split is on purpose. Parsing logic that lives in an extension goes stale in users’ browsers; parsing logic on the server improves for everyone at once.

The flow, step by step

1. Open a job page — a job board listing, a company careers page, an ATS-hosted application page, a job emailed to you as a link.
2. Click the SiviGen extension icon. The popup opens and immediately asks the page for its main text. You’ll see “Auto-captured” and the job title it found.
3. Read what it captured. The text sits in an editable box. Trim the boilerplate, delete the benefits section, or paste the description yourself if the page wouldn’t cooperate.
4. Press “Send to SiviGen”. This is the first moment any page content leaves your browser. (Opening the popup does check whether you’re logged in, which is a request to SiviGen and nothing else.)
5. SiviGen parses the job server-side — title, company, requirements — and runs the tailoring pipeline against your uploaded CV.
6. The application opens in SiviGen in a new tab. The popup, if you keep it open, shows live progress through the five pipeline stages: reading the job, matching your record, tailoring, verifying every claim, scoring.
7. Download the PDF straight from the popup, or open the application in the app to edit, review the score and read the claim-by-claim audit.

Total interaction: two clicks and a glance at the captured text.

Which job pages does it work on?

Any page it can read. There is no per-site scraper logic — no LinkedIn adapter, no Indeed adapter, no list of CSS selectors — and if Readability can’t isolate an article it falls back to the page’s plain text rather than giving up. So a small company’s own careers page and an ATS subdomain nobody has written an adapter for work the same way the big boards do.

Where it won’t work: browser-internal pages (chrome://…), the Chrome Web Store, and PDFs in the browser’s viewer — Chrome forbids extensions from running there. Some heavily scripted pages also defeat the article extractor. In each case the popup says it couldn’t auto-read the page and gives you the same editable box to paste into.

What does the extension actually send?

Specifics, rather than a privacy paragraph:

The request contains one field: the job text you approved. The content script is registered on all pages, and that’s worth being upfront about — it needs broad host access to read a job page on any domain, and so your login cookie reaches the SiviGen API. But registered is not active. It sits there holding a message listener and does nothing until the popup — which only opens when you click the icon — asks it for the current page’s text.

Authentication reuses the session you already have: if you’re logged into SiviGen in that browser, the extension is logged in. No API key to paste, no second account, no token stored in extension storage.

It remembers what you were doing, per tab

Small thing, disproportionately annoying when it’s missing. Close the popup while a job is sending or while your CV is being tailored, open it again, and it shows that tab’s real state — sending, ready, or needs attention — instead of resetting to a blank capture screen.

The record lives per tab in Chrome’s session storage. Navigate that tab to a different URL and it clears, because you’re looking at a different job now. Close the tab and it’s dropped. Close the browser and all of it is gone — session storage never touches disk.

What it doesn’t do yet

Plainly, because you’re deciding whether to install something:

• It isn’t on the Chrome Web Store. Today it ships as an unpacked build you load yourself via chrome://extensions → Developer mode → Load unpacked. There is no install link, and I’d rather say so than imply one exists.
• Chrome only. No Firefox or Safari build.
• It does not apply to jobs for you. It captures a job description and creates a tailored CV. Filling in application forms and pressing submit remain yours — deliberately, since that’s where the fabrication risk in “automated applying” tools lives.
• No right-click capture and no keyboard shortcut yet; the icon is the only entry point.

How does the job description become a tailored CV?

The extension is the front door. Behind it, your CV has already been parsed once into a fact graph of atomic, individually-addressable facts, and tailoring selects and rephrases from that graph — that’s how SiviGen tailors your CV.

The guarantee, though, is not that the writing step is incapable of overreaching. It’s what happens after it. A separate pass checks every line of the result back against your own record, labelling each claim supported (whether your CV says it outright or the inference follows safely from what it says), unsupported, or contradicted. A contradicted claim marks the result blocked and names the conflict, and the scoring that gates it is computed in code rather than judged by the model that wrote the text. The extension’s convenience never buys you a CV you can’t defend in the room. It buys you fewer tabs.

To see what the server sees before installing anything, run a posting through the free job description keyword extractor. If you’re weighing this against other tools, the test that settles it takes ten minutes: send any of them — this one included — a job description demanding a skill your CV never mentions, and read what comes back. And what SiviGen does covers the rest — scoring, the cover letter, the ATS-friendly PDF export.

FAQ

Is the SiviGen Chrome extension available on the Chrome Web Store?

Not yet. It currently ships as an unpacked build: you clone the repository, build the extension package, then load the output folder through chrome://extensions with Developer mode enabled. That means no one-click install link exists today. Web Store publishing is planned but not done, and the page you’re reading will be updated when it is.

Does it work on LinkedIn, Indeed and company career pages?

It works on any page it can read, because it uses Mozilla Readability rather than per-site selectors. There is no adapter for each job board, so nothing breaks when a board redesigns, and small company career pages and ATS-hosted listings work the same as the big boards. Browser-internal pages and in-browser PDFs are the exception — Chrome blocks extensions there.

Can the extension see my browsing history?

No. It reads the text of the page in the active tab, and only when you open the popup and it asks for it. No page content is transmitted until you press send — opening the popup does check whether you’re logged in — and the send request contains the job description text you approved, not the URL, not the page title, not any record of other pages. Closing the browser clears the extension’s per-tab state entirely.

Do I need to log in inside the extension?

No. It reuses the session cookie from your SiviGen login in the same browser, so if you’re signed into the web app you’re signed into the extension. If you aren’t, the popup shows a log-in link instead of a send button. If you’re signed in but haven’t uploaded a CV yet, it points you at onboarding first.

Does it apply to jobs automatically?

No, and that’s a design decision rather than a missing feature. The extension captures a job description and produces a tailored CV, which you review and submit yourself. Automated bulk applying pushes people towards sending documents they haven’t read, which is the exact failure mode the truthfulness audit exists to prevent.

---

Capture a job, get a CV built only from things you’ve actually done, and know which posting it was for. Start with SiviGen — upload your CV once, then let the extension handle the tabs. (If you call it a resume rather than a CV, it’s the same product; the tailoring doesn’t care what you name the file.)