Resume Parsing Architecture#
MonoForge includes a local parsing pipeline that extracts profile information directly from your existing resume without uploading documents to third-party AI cloud services:
- PDF Extraction: Backed by
pdf-parseextracting raw text layers and formatting. - DOCX Extraction: Backed by
mammothparsing document structure and hyperlinks. - Heuristic Classifier: Identifies technology keywords (TypeScript, Rust, Docker, Kubernetes), GitHub handles, and social links.
Supported File Formats#
- PDF documents (
.pdf) - Microsoft Word documents (
.docx)
Files are processed locally or through the self-hosted Next.js server endpoint (src/app/api/enrichment/resume/route.ts).
Mandatory Confirmation Gate#
Under the MonoForge architectural rules:
Note
**No parsed data may write to the profile store silently.**
When extraction completes, MonoForge opens the Confirm Gate Dialog (src/components/enrichment/ConfirmGate.tsx):
- Extracted technologies appear as individually selectable check boxes.
- Extracted social links show source URLs for your verification.
- Extracted bio statements allow inline editing before accepting.
You click Import Selected Items to write only the confirmed data into your profile store.