CLAUDE.md

This file provides guidance to Claude Code (claude.ai/code) when working with code in this repository.

What this repository is

A personal academic homepage for Xiangtai Li built on Jekyll, forked from RayeRen/acad-homepage.github.io. Deployed via GitHub Pages from the main branch at lxtGH.github.io. Day-to-day edits are content updates (publications, news, intro), not framework changes.

Local development

Requires Ruby + Bundler (Jekyll’s GitHub Pages stack).

bundle install              # one-time, installs github-pages gem
bash run_server.sh          # = bundle exec jekyll serve

Server runs on port 4001 (set in _config.yml, not the Jekyll default 4000 — the README is wrong about this). Livereload picks up edits to _pages/**, _includes/**, _sass/**, and assets/** automatically. _config.yml changes require a server restart.

Where content lives

The site is rendered from a single page (_pages/about.md) that include_relatives four content fragments — these are the files you almost always edit:

• _pages/includes/intro.md — bio, role, research interests, contact / hiring blurbs.
• _pages/includes/news.md — recent news (currently empty).
• _pages/includes/pubs.md — publication list. Hand-authored <li> entries with <strong> for the author’s name, a colored <span> venue/highlight tag, and trailing <a> links for code/project pages. Match this structure when adding new entries.
• _pages/includes/others.md — honors, awards, and other tail sections.

Project landing pages live under project/<name>/index.html (e.g. project/sa2va/, project/omg_seg/) and are linked to from pubs.md.

Site-wide layout lives in _layouts/default.html + _includes/ (masthead, sidebar, author profile, SEO, analytics). Styling is SCSS under _sass/. Touching these is rare — prefer content edits unless the user explicitly asks for a layout/style change.

Google Scholar citation pipeline

Per-paper citation counts are injected client-side, not at build time:

1. .github/workflows/google_scholar_crawler.yaml runs google_scholar_crawler/main.py daily (08:00 UTC) and on page_build. It uses the GOOGLE_SCHOLAR_ID repo secret and force-pushes gs_data.json + gs_data_shieldsio.json to a separate google-scholar-stats branch.
2. _includes/fetch_google_scholar_stats.html fetches that JSON via jsDelivr CDN (site.google_scholar_stats_use_cdn: true in _config.yml) and fills any element with class show_paper_citations, using its data attribute as the Google Scholar paper ID (citation_for_view=...).

To show a paper’s citation count anywhere in markdown:

<span class='show_paper_citations' data='SCHOLAR_PAPER_ID'></span>

The crawler script and google_scholar_crawler/ are excluded from the Jekyll build (see exclude: in _config.yml).

github_myprofile_updater/

Standalone script (update.py) that stitches the same _pages/includes/*.md fragments into a GitHub profile README. Note it reads homepage.md and pub_short.md which do not currently exist in _pages/includes/ — running it as-is will fail. Treat this as orphaned tooling unless asked to revive it.

Conventions when editing pubs.md

• <strong>Xiangtai Li</strong> is always bolded in author lists; co-first authors get a trailing *, with \* means equal contribution. noted at the top of the file.
• Venue tags follow the form <strong>[VENUE-YEAR]<span style="color:red"> one-line highlight </span> </strong>.
• Trailing links are plain <a href="...">Code</a> / Project Page / Model and Data, separated by spaces (not pipes inside the <li>).
• New “Recent Works” go to the top of that section; older ones drop into “Several Other Previous works”.