A lab homepage that also rebuilds a V2Ray subscription list

A lab homepage with a subscription feed attached

The repository exists to hold the homepage of the systems and networking lab at the School of Computer Science and Technology at Beijing Jiaotong University, and in practice it does more than that. The site files sit at the root: index.md with the _layouts/ directory and assets/ for the lab page and its resources, _config.yml, favicon.svg, faq-common.css, and a batch of PDFs and images including jiaogai1.pdf, jsuese.pdf, undergraduate-res.pdf, email1.png and email2.png. Alongside them are two workflow files under .github/workflows/, one that fetches and merges V2Ray subscriptions and one that takes a scheduled daily backup of a mentor homepage served on port 8530. A bat.bat file and v2ray.txt are also present, and GitHub reports JavaScript as the primary language. The published site is at https://fangvv.github.io/Homepage/ and the default branch is master.

Two workflows do work you would not expect from a homepage

The v2ray.yml workflow performs automatic fetching and merging of V2Ray subscriptions, and a later note records that this reached five sources being fetched, organised and published. The main.yml workflow is different in kind: it is a daily scheduled backup of a mentor homepage running on port 8530, which means the repository is also the backup target for a page it does not host. Both are ordinary Actions workflows and both carry a maintenance history. One note records that the API version in the GitHub Actions scripts was moved from v3 to v4, after which testing showed normal operation again, and that notes about updating a domestic Git service were removed at the same time. Another records that a large language model was used in the coding work, with the tools named as trae plus gpt5.5, at the point where the five source aggregation was working.

llms.txt exists because crawlers read this site

One root file is aimed squarely at machines rather than visitors: llms.txt, described as a site summary and navigation intended for LLM crawlers. The repository's most recent note, dated 15 September 2026, is about that file and about the two FAQ pages and the personal homepage. It records that a number of traps were embedded using steganography in those pages, llms.txt and the personal homepage, with the stated purpose of detecting large language models used to approach academics for filtering purposes, which is to say automated harvesting of contact details. Two earlier dated notes give the surrounding context. On 21 November 2025 a statement was added to the profile declaring the public research code, linking to the author's GitHub account, with a note that improper use of that code would be met with legal steps. On 5 February 2025 the Actions API update described above was carried out.

Four research pages sit next to the lab page

The repository is used as a container for several research groups, not only the lab itself. Edgecomp/index.html is a showcase page for edge computing and edge intelligence research, described in the contents table as a research display page. IndoorLoc/index.html is the indoor localisation research group. llm/llm.html covers on device large model research, and the same directory also contains the LLM material referenced elsewhere. Two FAQ pages have distinct audiences: faq.html is the FAQ for undergraduate research internships, and faq2.html is the FAQ for masters and doctoral admissions. Research experience sharing sits in exp.md with attachments under expfile/, and a papers/ directory is also present in the tree, though it is not itemised in the contents table. The table is not complete either, since favicon.svg, bat.bat and the PDF files are listed in the repository root but not described.

Anyone reusing the HTML has to strip the counter

The oldest note in the file, dated 25 October 2021, is a licensing condition in practice rather than in licence form. Anyone who copies code from the site is asked to remove the statcounter counter code embedded in the HTML pages before using them, and the reason given is specific: leaving it in disturbs the site's own statistics while the copy is being debugged. The same note says that copying the site's text or images is not welcomed, and that such content belongs to the author or the lab. There is no LICENSE file in the repository root, so nothing here grants reuse rights in writing. Two PDF documents in the root, jiaogai1.pdf and undergraduate-res.pdf, suggest the site also carries recruitment material, which makes that note about copying worth taking literally.

Two notes about reaching GitHub from inside China

Two dated notes are about connectivity rather than about the site. The first, from 13 November 2023, reports that code could not be updated on GitHub and gives the fix as a proxy command:

git config --global https.proxy 'socks5://127.0.0.1:10808'

The note is careful about failure modes: writing the value incorrectly has no effect, and neither does setting the proxy while nothing is listening on port 10808. The second, from 2 June 2024, is about page weight rather than network access, recording that images on the pages were compressed substantially with TinyPNG with domestic access speed in mind. Both notes are the kind of thing a README usually drops, and both are still useful: one is a git configuration you can paste, the other is a reminder that image weight is a first order problem for a site read from a slow link.

Speed testing is a GitHub issue thread

The 13 November 2023 note also handles a different failure, inaccurate speed test results, and its answer is a link to an issue on CloudflareSpeedTest, plus a suggestion to search for a VPS looking glass alternative when you need another speed test site, with 10mb being the size suggested as appropriate. That is an unusually honest admission that a third party tool was producing bad numbers, and the fix is to change the measuring instrument rather than to argue with the result. Read together with the two connectivity notes, the update history from 2021 to 2026 covers four separate operational concerns: a counter that must be removed before reuse, a proxy that must be configured correctly, a measurement that must be done with the right tool, and images that must be compressed. None of it is about the site's design.

Editorial conclusion

Treat this repository as a working lab site with the automation of a small operations team attached to it: one workflow aggregates V2Ray subscriptions and another takes a daily backup of a page served on port 8530, which is a different job from hosting a homepage. It suits anyone who wants a static lab site with the same file layout, or who reads Chinese notes on the problems of running a site and a subscription feed from inside China. It does not suit a project that needs a licence, releases or a documented build, since none of those exist here. Before copying any of it, strip the statcounter block as the README asks, and expect the workflows to need their own credentials rather than working out of the box.

Frequently asked questions

Official sources

1. Official documentation
2. Official README
3. Project repository