菜单
Use when the user asks to fetch, crawl, collect, scrape, or refresh conference paper candidates from CVPR/CVF pages, RSS/Atom feeds, manual paper lists, or topic filters before reading or building the site.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Use when Codex needs to generate abstract-only Obsidian Markdown notes from one or more data/papers/*.json files by reading arXiv HTML abstract metadata when available, without reading the full paper. Trigger for requests like "Use $conference-papers-readsimply 简读 data/papers/xxx.json", "批量简读 data/papers/*.json", or "generate simple abstract md notes"; this skill is independent from website note JSON and dist build.
Use when building or maintaining a searchable HTML reading site from conference papers, CVPR/CVF virtual pages, RSS/Atom feeds, manual paper lists, or topic-specific paper collections such as VLA, Humanoid, locomotion, or loco-manipulation.
Use when Codex needs to read a conference paper from data/papers/*.json and generate a standalone Obsidian-compatible deep-reading Markdown note under data/_inbox, with online images by default and optional localization only when needed. Trigger for requests like "Use $conference-papers-read2md 精读 data/papers/xxx.json", "根据 paper JSON 生成 Obsidian md", or "read conference paper JSON to Markdown"; this skill is independent from the conference-papers website note JSON and dist build.
Use when the user asks to maintain an existing conference paper website by adding papers, merging topics, deduplicating records, enriching metadata, fixing related-work links, or rebuilding after updates.
Use when the user asks to read selected conference papers, generate high-quality paper notes, convert draft notes into final notes, or populate structured note JSON for the conference paper website.
Use when the user asks to build, rebuild, render, preview, or regenerate the static HTML website, topic pages, paper pages, assets, or search index from existing conference paper data.
| name | conference-papers-fetch |
| description | Use when the user asks to fetch, crawl, collect, scrape, or refresh conference paper candidates from CVPR/CVF pages, RSS/Atom feeds, manual paper lists, or topic filters before reading or building the site. |
Use this thin entrypoint for the fetch stage of the conference paper website. The implementation lives in ../conference-papers; do not duplicate scripts here.
data/.../conference-papers/config/user-config.json; allow a site-local config/user-config.local.json to override conference, year, topics, source URL, and paths.../conference-papers/references/source-adapters.md.python3 ../conference-papers/scripts/add_papers.py --draft
python3 ../conference-papers/scripts/add_papers.py --conference CVPR --year 2026 --topic VLA --url "https://cvpr.thecvf.com/virtual/2026/papers.html" --data data --draft
python3 ../conference-papers/scripts/crawl_conference.py --conference CVPR --year 2026 --url "https://cvpr.thecvf.com/virtual/2026/papers.html" --topics VLA Humanoid locomotion --out data/conferences/cvpr-2026.json
python3 ../conference-papers/scripts/enrich_arxiv.py --data data
Fetch should update data/papers/*.json and data/conferences/*.json. Report how many papers were added or merged.