# airport-access: the shortlist behind a Chinese-language VPN provider guide

> airport-access is a two-file content repository that publishes a short list of VPN providers, with a daily node availability table and a running log of changes. It suits readers who want to see how such a list is maintained and when it was last checked, and it is of little use outside its target market.

**chatgpt-helper-tech/airport-access** — 机场推荐 | 科学上网 · 魔法上网 · 梯子推荐 · VPN推荐 | 精选稳定机场

- Repository: https://github.com/chatgpt-helper-tech/airport-access
- Website: https://gptvpnhelper.com
- Stars: 756 · Forks: 76
- Language: Unknown
- License: not declared
- Published: 2026-10-21 · Updated: 2026-10-21 · Language: en
- Canonical page: https://hysenlabs.com/projects/chatgpt-helper-tech-airport-access

## A repository with two markdown files and no code

The repository contains exactly two top level entries, README.md and daily-news.md. That is the whole thing. There is no build script, no package manifest, no test suite, and no source directory, so this is a publishing project rather than a software project.

The README is a long-form recommendation article in Chinese, and the repository description is likewise Chinese, listing the topic as provider recommendations for scientific internet access and VPN recommendations. The homepage given for the project is gptvpnhelper.com.

The README itself carries a content date of 2026-10-15, which is separate from the repository's last push on 2026-10-20. That gap matters for anyone treating the numbers inside as current: the article's figures were assembled on the 15th, and the repository was touched on the 20th.

Repository topics are clash, shadowsocks, ssr, v2ray and vpn, which tells you the list is aimed at readers who already know those client names. The language of the target audience is not incidental, it is the whole frame, since the provider availability claims, the geographic restrictions and the referral links are all written for one market.

## The daily node check table and what its percentages mean

The most concrete part of the repository is a table of node availability observations. Each row names a provider, the client used for the test, the date of the last check, whether the service was reachable, the total node count, the reachable node count, the ratio for that run, a historical ratio, and a stability rating.

For the check date of 2026-10-15, the recorded figures are: BoostNet on its official client with 44 of 45 nodes reachable, 97.78 percent against a historical 95.56 percent, rated high stability; WgetCloud on Clash Verge with 29 of 29, 100 percent in both columns, rated high; YouTu on its official client with 49 of 51, 96.08 percent against 93.81 percent historically, rated high; XiaoMiFeng with 27 of 32, 84.38 percent against 86.05 percent, rated medium; U1S1 with 60 of 68, 88.24 percent against 92.11 percent, rated high; TAG with 301 of 304, 99.01 percent against 98.92 percent, rated high; and LingMao with 45 of 50, 90 percent in both columns, rated medium.

The note under the table defines the current ratio as reachable nodes over total nodes, and the historical ratio as a long term running average. Six of the seven providers were checked with their own official client and only WgetCloud was checked with Clash Verge, which is a methodological inconsistency worth noticing when comparing rows. LingMao was added on 2026-10-12, so its historical column has barely started accumulating. The table also states that the data reflects connectivity at the moment of observation only.

## Seven providers with different protocols, prices and client restrictions

A comparison table sets out the seven recommended providers across line protocol, AI service coverage, streaming coverage, device limits and caveats.

BoostNet is listed at 49 yuan per month for 200GB, using direct IEPL lines on the three main carriers plus Anytls and its own client, with five simultaneous devices on the regular plans and use limited to mainland China with Xinjiang excluded. WgetCloud starts at 297 yuan for 90 days, roughly 99 yuan per month averaged, on Guangzhou BGP, IEPL and Trojan plus Shadowsocks, with three, four or five online IPs depending on the tier, and its registration link itself requires the proxy environment to open.

YouTu is priced at 39 yuan per month for 150GB with a default of five devices per IP, and bans BT, P2P and PT traffic. XiaoMiFeng is 20 yuan per month for 120GB with unlimited devices and a Shenzhen entry on Deep HK IEPL with Trojan, and currently requires its own client. U1S1 is also 20 yuan per month for 120GB on IPLC, IEPL and BGP with Shadowsocks, and the guide flags a subscription reset rule.

TAG is listed at 162 yuan a year on its Special plan and 185 yuan per quarter on Bronze, with 330 or more routes across more than 100 countries, up to ten devices on the Bronze, Silver and Gold plans, five on Special and thirty on Team, and a restriction to mainland China only. LingMao is 25 yuan per month for 150GB over full IPLC with VLESS, without speed multipliers or client count limits, but it does not support Clash and points iOS users at Nextin.

## Why the list is kept short and the change log that proves it moves

The README argues for keeping the list deliberately short, at seven providers rather than a catalogue, on the grounds that the industry became noticeably more volatile from April 2026, with providers seeing long entrance outages, unreachable support and unclear announcements. The stated priority for the reader is a provider with dependable line quality, entrance maintenance and support response, not the lowest price.

Three rules are named: control the total, track over the long term and adjust dynamically, and prefer stability to low price. The change log then shows the adjustment happening.

On 2026-10-15 the WgetCloud 90-day tier prices and configuration were updated along with traffic quotas and a 30-day reset rule. On 2026-10-12 LingMao was added as an all-IPLC and VLESS provider with a late peak YouTube 4K measurement, taking the list to seven. On 2026-09-25 the XiaoMiFeng recommendation was restored. On 2026-09-23 WgetCloud was restored in the team and enterprise stability slot. On 2026-08-24 TAG was added and a previous team option was removed. On 2026-08-01 WgetCloud reopened its 90-day plan. On 2026-07-13 node speed and streaming measurements were refreshed for u1s1, BoostNet and YouTu, and on 2026-05-02 the u1s1 plan structure changed.

That record is the useful part of the repository for a critical reader. It shows providers entering and leaving, and it shows prices and quotas moving, which is exactly what a static recommendation page cannot show.

## The referral links are the business model, not a footnote

Every provider entry in the quick choice table and the comparison table carries a registration link with a referral code in the query string, and the guide explicitly tells the reader to click through after choosing. The stated advice is to buy a monthly or short cycle plan first, confirm that the local carrier, the commonly used nodes, the AI tools and the streaming services all work, and only then commit to a longer plan, on the grounds that this keeps the trial cost low.

That is a reasonable piece of advice on its own terms, and it is also the standard funnel for an affiliate site. The repository therefore mixes two different kinds of statement, and a reader has to separate them by hand.

On one side are the mechanical observations, such as node counts and the last check date, which come from the site's own daily testing. On the other side are the provider descriptions, including claims about which streaming services are reachable, which AI services work on which nodes, and how many routes and countries a provider has. Those come from the providers and are repeated by the site.

The guide itself offers one honesty check when comparing providers: it asks the reader to weigh stability, line quality, entrance maintenance, support response, purchase threshold and use case together, and it says plainly that a reader who only wants the cheapest option will not find what they need here. It also notes that TAG makes no guarantee for TikTok, and that several providers restrict use to mainland China.

## What this repository cannot tell you

Several things a reader might want are simply absent, and the absence is informative.

There is no method description for how a node is tested. The table records how many nodes answered, but the README does not state how connectivity is probed, how long a node has to respond, whether a node returning the wrong content counts as available, or whether the historical average is weighted by time or by runs. Without that, the ratios are comparable to each other but not to any measurement you make yourself.

There is also no machine readable data. The availability figures live in a markdown table in the README, so the values cannot be consumed by a script or tracked automatically. The daily-news.md file at the top level suggests a running news feed, but the README excerpt does not describe its format or how it relates to the tables.

The pricing is quoted in yuan per month or per quarter with no tax, currency conversion or long-term discount treatment, and the reset and quota rules differ enough between providers that a direct price comparison across rows is misleading. WgetCloud is deliberately positioned at the expensive end for stability, TAG prices by year and quarter rather than by month, and two of the seven do not support a general Clash client at all, which changes the real cost if you already run one.

Finally, the repository has no tagged releases and no stated licence, so there is no version marker to pin and no clear permission for redistribution of the content or the linked commercial pages.

## Conclusion

airport-access is worth reading if you want to see the structure behind a curated provider shortlist, including how availability is sampled and how the list changes over time. It is not usable as an engineering dependency, since the repository contains only README.md and daily-news.md and no code, and it targets users on the Chinese mainland, where several of the listed providers explicitly do not sell. Before relying on any figure in it, check the last check date in the availability table against the current day, and treat every price, protocol and streaming claim as the site's own reporting rather than as independently confirmed data.

## FAQ

### What is the airport-access repository?

It is a content repository holding a Chinese-language guide that recommends a short list of VPN providers, with a daily node availability table and a dated log of changes. Its only top level files are README.md and daily-news.md, so there is no code involved.

### How does airport-access measure node availability?

The table records the reachable node count over the total node count for a given day, plus a long term running average. The README states that each provider is tested with the client listed in its row, with Clash Verge used for WgetCloud and official clients for the rest.

### Which providers does the airport-access shortlist currently include?

Seven: BoostNet, WgetCloud, YouTu, XiaoMiFeng, u1s1, TAG and LingMao. The README says the list is intentionally short and that providers are added or removed as line stability, entrance maintenance, support response and plan changes are observed.

### Which providers in the airport-access list do not work with Clash?

LingMao is listed as supporting its own official client only, with Clash not supported for now, and iOS users directed to Nextin. XiaoMiFeng is also described as currently requiring its official client, though the note says to defer to the provider's own site for the current rules.

## Sources

- [chatgpt-helper-tech/airport-access on GitHub](https://github.com/chatgpt-helper-tech/airport-access)
- [Issues](https://github.com/chatgpt-helper-tech/airport-access/issues)
- [Project website](https://gptvpnhelper.com)
- [README](https://github.com/chatgpt-helper-tech/airport-access/blob/main/README.md)

---

Hysen Labs editorial analysis, written from the project's own repository and release notes. Cite the canonical page: https://hysenlabs.com/projects/chatgpt-helper-tech-airport-access
