PaperPrism

PaperPrism turns your arxiv PDF downloads into a searchable, auto-classified local library — without sending a single paper to the cloud. ━━━ What it does ━━━ • Watches arxiv downloads and archives each paper automatically. • Extracts title, authors, abstract, and publication date from the PDF and the arxiv API. • Classifies papers into multi-dimensional labels (topic, task, methodology, venue) using your own LLM — running on your machine or via your own API key. • Auto-generates 2–5 concise tags per paper on ingest. • Lets you batch-select papers and synthesise them into a named Topic with a one-liner summary. • Built-in Dashboard to browse, full-text search, filter by tag / topic / dimension, view PDFs inline, and bulk-import an existing folder of papers. ━━━ Privacy ━━━ All data stays on your device. PDFs live in a hidden local vault; metadata lives in a local SQLite database. The extension communicates only with your own machine (127.0.0.1). No telemetry, no analytics, no cloud sync. The optional LLM classification step calls the provider you choose, governed by that provider's own privacy policy. You can also use a fully local model to keep everything offline. ━━━ How it works ━━━ PaperPrism consists of two parts: 1. This Chrome extension — the capture layer and UI. 2. A lightweight local Agent (Python, ~15 MB installed) that handles storage and classification. Install the Agent with a single command: uv tool install paperprism-agent paperprism-agent serve Then click the PaperPrism toolbar icon and follow the 4-step setup wizard. ━━━ Open source ━━━ MIT-style Apache-2.0 license. Source: https://github.com/MrMao007/PaperPrism