Local Source Ingestion
Currently running as a real personal workflow (agents, markdown files, scripts). A written explainer post covering the current setup is planned, and it's a candidate for a short walkthrough video too.
Personal research leaves a trail of scattered sources — PDFs downloaded into random folders, URLs bookmarked or left open in browser tabs, spreadsheets exported from one service or another. Nothing unifies that into one queryable, organized archive without either requiring a cloud service or bolting on AI features that aren’t actually needed for the job.
This pipeline ingests URLs and PDFs, extracts useful metadata (title, word count, source structure) as each item comes in, and stores everything in a structured local database that’s browseable by type, date, or source — turning a scattered stream of “things I meant to come back to” into an actual retained, organized archive.
It’s designed as the ingestion layer underneath Content Triage App: this captures raw sources into a structured store; triage handles the decision of what to do with each one. No AI/LLM features by design — this is about getting things captured and organized, not summarized.