v0.1.0-alpha — First Public Release

Paper Workflow v0.1.0-alpha

First public release. The core pipeline works end-to-end.

Highlights

• Multi-source search: arXiv + Semantic Scholar with deduplication
• Full ingest pipeline: PDF download, PyMuPDF parsing, chunking, embedding, FAISS + SQLite indexing
• Hybrid retrieval: vector + full-text + RRF ranking
• Structured single-paper analysis with evidence references
• 5 Claude Code skills, 1 batch ingest agent
• Zero CC source modification
• 140+ tests

Current Limitations

• Best with arXiv-quality PDFs. Scanned or complex layouts may parse poorly.
• No citation graph or reference extraction yet.
• Multi-paper comparison and lit review generation not yet implemented.
• Embedding model must be configured via .env.

Install

git clone https://github.com/Minazuki02/paper-workflow.git
cd paper-workflow && pip install -e ./backend
cp .env.example .env && claude