paper-scribelisted

# PaperScribe ## Trigger Phrases This skill should trigger naturally without requiring the user to say `paper-scribe`. Treat these as direct requests to use this skill: - `读一下这篇论文` - `读一下这个` followed by an arXiv link, PDF path, or paper title - `读一下这篇` followed by paper context - `帮我读这篇 paper` - `论文笔记` / `深度读论文` / `分析这篇论文` - Any arXiv link or local PDF path when the surrounding request is about reading or making notes for a paper PaperScribe deep-reads one academic paper and creates an Obsidian Markdown note. This skill is optimized for English arXiv-style papers, text-based PDFs, Zotero local attachments, and section-by-section Chinese reading notes. ## Hard Requirements Before doing any paper work: 1. Load `config.local.json` from this skill directory. If missing, run `python scripts/init_config.py` in the Python environment chosen by the user, or ask for `note_output_dir`, `pdf_download_dir`, optional Conda/Python settings, and Zotero mode: skip, auto-detect, or manual paths. 2. Require `note_output_dir` and `pdf_download_dir`. Stop if either is empty or invalid. 3. Use `assets_dir` from config if present; otherwise use `{note_output_dir}/assets`. 4. If `python.conda_env_name` is set, run bundled scripts as `conda run -n {conda_env_name} python ...` unless the user overrides it. If `python.executable` is set and no Conda env is set, use that executable. Otherwise use `python`. 5. Zotero paths may be empty, auto-detected, or user-provided. If configured paths are inv