GitMatch scans real dependency files — not just bios — and returns a precise match score against any job description. Built for hirers who need signal, not noise.
The analysis pipeline goes deeper than any profile scraper. It reads actual code — the packages you installed, the tools you depend on.
GitMatch pulls your public GitHub profile and top repos via the GitHub REST API. Read-only scope — no write access requested.
The engine recursively crawls top repositories, descending into nested subdirectories (like client, server, and monorepo workspaces) to parse package manifests and locate test environments or CI configs.
Paste any job posting. The JD parser pulls out the role, required experience years, and technical skills — no special formatting needed.
Matching runs inside an isolated Worker Thread using a dual-engine score: Job Fit (70% weight) and Engineering Quality (30% weight across 6 hygiene axes) are combined into a final rating without locking the UI.
Review rich dashboard lists, compare up to 29+ candidate matrices side-by-side, and generate beautiful, layout-budgeted multi-page vector PDF reports completely offline.
Every feature exists to remove one more layer of gut-feel from the hiring process.
Scans nested directories (like client, server, frontend, packages) for package files, Dockerfiles, and test suites to accurately evaluate monorepos and MERN stacks.
Parses Job Descriptions, normalizes skill priorities (weights sum to 100), and scores candidate compatibility scaled by project confidence.
Grades code bases across 6 distinct hygiene dimensions: Documentation (README), tests, CI/CD pipelines, modular layout, cadence, and complexity.
Recruiter dashboard comparing shared/exclusive skills, confidence counts, and engineering hygiene side-by-side. Dynamic layout wraps for up to 29+ profiles.
Generates beautiful candidate, screening summary, and comparison sheet PDFs completely offline (using PDFKit) with page budget safeguards.
Delegates AST parsing, score math, and Git graph processing to a background worker thread, keeping the Electron desktop UI butter-smooth.
Persisted local screening history with instant keyword search, report reopening, and secure double-confirmation delete overlays.
Seamless transition between high-contrast dark theme (ideal for developers) and a soft cream light mode styling system.
Official GitHub OAuth 2.0 flow token storage using Electron's native credentials isolation store. No external server stores your keys.
Every layer is chosen deliberately — not because it's trendy, but because it's the right tool for what GitMatch needs to do.
| Desktop | Electron 29 | Context Isolation · IPC |
| Frontend | React 18 + Vite | React Router DOM |
| Styling | Vanilla CSS | Custom design system |
| Icons | Lucide React | Accessible, scalable |
| Backend | Node.js + Express | OAuth server |
| GitHub API | @octokit/rest | REST + GraphQL |
| Concurrency | Worker Threads | Node.js built-in |
| Storage | electron-store | Local token cache |
| PDF Engine | PDFKit | Offline vector layout |
All GitHub OAuth token exchanges happen inside the secure Node.js main process. The renderer — and therefore any injected script — never sees credentials.
Renderer cannot access Node.js APIs directly. Only the strictly typed electronAPI surface is exposed via the preload script.
Client ID and Secret are loaded at runtime from .env via dotenv. They never appear in the bundled renderer code or version control.
The preload bridge exposes exactly the channels GitMatch needs — nothing more. No broad require() access from the renderer side.
Free desktop app for Windows. Connect your GitHub OAuth App in 5 minutes and score your first candidate immediately.