Do the channels for Linux apps let AI crawlers in?
On 2026-11-11 we read the robots.txt of the common channels for Linux apps one by one and judged them for seven User-Agents; the rule is that a named group wins, and without one the * group applies. This only shows how robots.txt is written. It does not show that those crawlers actually came, or that AI cited these pages (we have no evidence of citations: not verified).
| Channel | GPTBot | OAI-SearchBot / ChatGPT-User | ClaudeBot | PerplexityBot | Google-Extended | Bingbot |
|---|---|---|---|---|---|---|
| Flathub (flathub.org) | Blocked | Allowed | Allowed | Allowed | Blocked | Allowed |
| Snap Store (snapcraft.io; only login and publisher dashboards are blocked) | Allowed (only some paths restricted) | Allowed (only some paths restricted) | Allowed (only some paths restricted) | Allowed (only some paths restricted) | Allowed (only some paths restricted) | Allowed (only some paths restricted) |
| AUR (aur.archlinux.org; search pages and pkgbase are blocked) | Allowed (only some paths restricted) | Allowed (only some paths restricted) | Allowed (only some paths restricted) | Allowed (only some paths restricted) | Allowed (only some paths restricted) | Allowed (only some paths restricted) |
| AppImageHub directory (appimage.github.io; only a Sitemap line) | Allowed | Allowed | Allowed | Allowed | Allowed | Allowed |
| AppImageHub store (www.appimagehub.com; unreachable) | Not verified | Not verified | Not verified | Not verified | Not verified | Not verified |
| GitHub (repository pages and releases pages) | Allowed (only some paths restricted) | Allowed (only some paths restricted) | Allowed (only some paths restricted) | Allowed (only some paths restricted) | Allowed (only some paths restricted) | Allowed (only some paths restricted) |
How to read this table: Flathub is the only one that names and blocks training crawlers (GPTBot, Google-Extended and others) while allowing search crawlers; Snap Store and AUR name no AI crawler and only block pages with no content such as login and search; GitHub adds Crawl-delay: 1 for AI crawlers. For how to write robots.txt see How to write robots.txt so AI crawlers can get in.
Conclusion: none of these channels stops AI search crawlers from reading app detail pages, so the question is not whether they can get in but what is on the page. Next, the fields readable on each page.
Which AppStream metainfo fields become a Flathub app page?
Flathub pages are generated from the AppStream metainfo (an XML file) you submit. We fetched the GIMP page on Flathub (2026-11-11, HTTP 200, about 223 KB) and its public API; the fields line up as follows. The requirements come from the Flathub metainfo-guidelines docs (viewed 2026-11-11).
| metainfo field | What you see on the Flathub page | Official requirement and how to write it |
|---|---|---|
| name | Page title: Install (name) on Linux | Flathub | Must not infringe trademarks |
| summary | The meta description and the subtitle under the title | One sentence saying what it does |
| description | Page body, written directly in the HTML | Required; only p, ol, ul, li, em and code are supported; the first paragraph answers what it is and who it is for |
| screenshots | Screenshots and captions | Graphical apps need one or more; console apps need screenshots too to appear in the site's search results |
| releases | “Changes in version …” and the time | A releases tag is required; the description uses the same format as description |
| url | The Links area: website, help, bug reports, source, contribute | At least a homepage type is required |
| developer | The developer name under the title | Required, with a reverse-DNS id |
| keywords, categories | The Tags area and keywords and categories in the public API | Write the words users search for |
<component type="desktop-application">
<id>com.example.ExampleApp</id>
<name>ExampleApp</name>
<summary>One sentence: what it does and who it is for</summary>
<developer id="com.example"><name>Example Co</name></developer>
<description>
<p>First paragraph answers: what is it, for whom, what does it replace.</p>
</description>
<url type="homepage">https://example.com/</url>
<url type="bugtracker">https://github.com/example/app/issues</url>
<screenshots>
<screenshot type="default"><image>https://example.com/shot1.png</image><caption>Main window</caption></screenshot>
</screenshots>
<releases>
<release version="2.3.0" date="2026-11-01"><description><p>What changed.</p></description></release>
</releases>
</component>
The page has no SoftwareApplication structured data (only Organization, WebSite and BreadcrumbList): crawlers read the title, the meta description and the server-rendered body. So the summary and the first paragraph of the description are the sentences others will quote; make them explain the app as clearly as the first paragraph of a website would.
One reminder: Flathub's requirements page says console software is generally not accepted (except Flatpak or Flathub tooling), while the metainfo guidelines also mention screenshots for console apps. Whether an app can be listed depends on Flathub's rules at the time; we make no judgment.
What can be read on Snap Store, AUR and AppImageHub pages?
On all three channels the page body is server-rendered; the differences are in structured data and where the description comes from. On 2026-11-11 we tested one page on each.
| Channel | title and meta description | Structured data | What the body shows |
|---|---|---|---|
| Snap Store (snapcraft.io/vlc) | The title is Install (name) on Linux | Snap Store; the description is a template sentence plus the summary | SoftwareApplication: name, description, version, file size, license, operatingSystem, publish date | Package name, license, last updated, website, source, bug-report link, full description |
| AUR (aur.archlinux.org/packages/yay) | The title is AUR (en) - package name; there is no meta description | None | A one-line Description, Upstream URL, Keywords, license, maintainer, votes, popularity, first-submitted and last-updated times, dependencies |
| AppImageHub (appimage.github.io/Inkscape/) | The title is name – AppImages; the description is one sentence used across the whole site, not the app's own | WebPage only | The full app description, website, bug-report and help links, AppImage usage notes |
An AUR description is a single sentence, shown on the AUR page and in pacman search results, so write it as one “what it does, for whom” sentence. The AppImageHub page description is generic and the app description sits in the body, so the first paragraph of that description matters most. We did not read the official field docs for these three channels (not verified).
What should the README and the project website say?
Of all the channels, the only ones you fully control are your own website and your GitHub repository. A GitHub repository page's title is “owner/repo: description”, its meta description is the repository description plus a fixed template sentence, the README text is written straight into the HTML, and there is no structured data (checked on the Jguer/yay page on 2026-11-11). So the repository About description and the first paragraph of the README are the two places AI reads most easily; see How should a GitHub README be written so AI quotes it? for how to write them, and How do open-source projects get recommended by AI? for the general open-source project pages (license, contribution guide).
| Location | What goes in | Why |
|---|---|---|
| GitHub repository About description | One sentence: what it does, for whom, which distributions it supports | It is the source of both the title and the meta description |
| First paragraph of the README | The same sentence expanded: what it is, how it compares, how to install it | It is in the HTML, so crawlers need no script |
| Website download page | Install commands for each channel (flatpak, snap, AUR, AppImage), supported distributions and architectures, current version | Each channel's page format is set by someone else; the website is where you can write it all |
| GitHub releases and changelog page | What changed in each version; dates and version numbers consistent with each channel | Version notes on channel pages may fall out of sync; keep the latest on your website |
How do you make a Linux app easier to recommend, step by step?
1. First make sure your website and repository do not block search crawlers
Use AI crawler check to look at your website's robots.txt.
2. Write a good summary and description in the AppStream metainfo
The Flathub page, the public API and app store clients all read it. Make summary one sentence and the first paragraph of description answer “what is it, who is it for”; adapt the example above.
3. Fill in screenshots, releases and url
In releases, write what changed in every version; url needs at least a homepage, and a bugtracker is a good addition. For how to write your website's changelog page see Do changelog pages help AI search?.
4. Keep the descriptions on other channels in step
Snap's summary, AUR's Description, AppImageHub's blurb and the repository About description should all be the same sentence at different lengths.
5. Add SoftwareApplication structured data to your website
Write Linux in operatingSystem; for how to write it see Should a SaaS add SoftwareApplication structured data?.
6. Test again after a few weeks with the same questions
Ask ChatGPT and Perplexity “what apps on Linux do X”, three times per question, and note whether you are mentioned.
FAQ
Will ChatGPT cite a Flathub page?
We found no official statement and no evidence of actual citations (not verified). What we can confirm: on 2026-11-11 Flathub's robots.txt places no restriction on OAI-SearchBot, ClaudeBot or PerplexityBot, and the app page's summary and description are written straight into the HTML.
Flathub blocks GPTBot; can ChatGPT search still read it?
According to OpenAI, GPTBot is a training crawler and ChatGPT search uses OAI-SearchBot; on 2026-11-11 Flathub's robots.txt does not name OAI-SearchBot, so the * group applies and it is allowed.
Which part of the AppStream metainfo deserves the most time?
The summary and the first paragraph of the description. The Flathub page's meta description is the summary and the body is the description; both are the sentences others are most likely to copy when they cite you.
Is one sentence enough for an AUR page description?
The AUR Description is a single sentence that appears on the page and in package manager search results. Write it as “what it does, for whom” and put longer explanations in the GitHub README and your website.
Where should an AppImage be published?
We only verified the AppImageHub directory page and its robots.txt, and did not read its listing rules (not verified). Wherever you publish, state the download method and supported distributions on your website and the GitHub releases page.
Should I publish on Flathub, Snap and AUR all at once?
The channels depend on which distributions your users run, not on AI. When you use several, keep names, descriptions and version numbers consistent; where they differ, AI may mix them up when answering.
Does a Linux app's website still need structured data?
Yes, we suggest adding it. The Flathub page we fetched has no SoftwareApplication data while Snap Store does; adding it to your website gives the name, system and version one clear machine-readable source. See the steps above for how to write it.
Sources
- Flathub · flathub.org/robots.txt (Read 2026-11-11)
- Flathub · Example app page: GIMP, and flathub.org/api/v2/appstream/org.gimp.GIMP (Read 2026-11-11)
- Flathub · MetaInfo guidelines (checked 2026-11-11)
- Flathub · Requirements (checked 2026-11-11)
- Snapcraft · snapcraft.io/robots.txt (Read 2026-11-11)
- Snapcraft · Example snap page (checked for what the HTML contains) (Read 2026-11-11)
- AUR · aur.archlinux.org/robots.txt (Read 2026-11-11)
- AUR · Example package page (checked for what the HTML contains) (Read 2026-11-11)
- AppImageHub · Example app page; appimage.github.io/robots.txt (www.appimagehub.com was unreachable, not verified) (Read 2026-11-11)
- GitHub · github.com/robots.txt and an example repository page (Read 2026-11-11)
- kagentx evaluations: the data on the case study pages; each example gives its date.