ApplyOS: A Local-First Command Center for the Job HuntApplyOS

Applying to jobs means re-typing the same 40 fields and re-answering the same 5 questions hundreds of times, then losing track of where you applied. ApplyOS removes that loop with four local-first parts: a Vault that stores every field and snippet once (with {{company}}/{{role}} variables that resolve when you copy), a Tailor that scores a job description offline and drafts tailored bullets, a Tracker for the full pipeline, and a one-click autofill extension. Job-hunt data is sensitive, so there is no backend state at all, only local IndexedDB with JSON export.

Autofill is heuristic rather than per-site: every input is scored by its surrounding signals (label, placeholder, name, aria-*, autocomplete) against an ordered pattern table, so it works on ATSs it has never seen. For React-controlled forms it writes through the native prototype value setter and dispatches input/change/blur events, so frameworks register the change like a keystroke. Auto-capture parses JSON-LD JobPosting first (then URL slugs, then the <h1>) and queues captures in a chrome.storage outbox that drains into IndexedDB the next time the app opens. Passwords are encrypted with AES-256-GCM under a PBKDF2 key that is non-extractable, held in memory only, wiped after 5 idle minutes, and never synced to the extension.

Fill your vault once (about 15 minutes), paste a job description for an offline match score, then fill an application in one click and have it logged to your tracker, with an amber "already applied 4 days ago" warning if you've been here before. The dashboard shows which profile and which portal convert best.

“

Long-form answers stay manual on purpose: you copy from the vault with variables resolved, and nothing is auto-submitted. Next: resume-PDF parsing to populate the vault, fuzzy dropdown filling for degree and country pickers, and a Firefox port (MV3 is nearly compatible).