At build time, Shirone generates static outputs for feed readers, search engines, and AI tools. They all use the same site configuration and post data, with no runtime server required.

Output Catalog

URL	Format	Purpose	Generation
/rss.xml	RSS 2.0	Feed readers and aggregators	Always generated
/atom.xml	Atom 1.0	Atom-compatible clients	Always generated
/sitemap-index.xml	XML Sitemap	Search-engine discovery	Generated by Astro Sitemap
/robots.txt	Plain text	Crawler rules and Sitemap declaration	Always generated
/llms.txt	Markdown	Site index for AI tools	llms.enable: true
/llms-full.txt	Markdown	Full public-post text dump	llms.enable and llms.generateFull are true

All URLs use siteConfig.site (or the build context's site) as their absolute origin and respect Astro's base setting when deployed under a subpath.

---

RSS and Atom

Both feeds receive post data from getFeedPosts(). They include publicly published posts with titles, publication dates, descriptions, links, and rendered HTML content.

• RSS is generated with @astrojs/rss; its language comes from siteConfig.lang.
• Atom is generated by Shirone's buildAtomXml(); its author comes from profileConfig.name.
• Relative media references are converted to accessible site URLs in feed content.
• Drafts are excluded from production builds; they remain available for development preview.
• Encrypted posts remain as feed entries but are sanitized: titles receive a lock marker and content becomes a password-protection notice. When home excerpts are hidden, the description also uses a secure notice.

Feed readers such as Feedly and Inoreader can subscribe directly:

https://example.com/rss.xml
https://example.com/atom.xml

---

Sitemap and robots.txt

The Astro Sitemap integration generates sitemap-index.xml from pages emitted by the build. Articles, standalone pages, and pagination routes appear according to whether their corresponding pages are generated in that build.

src/pages/robots.txt.ts emits:

User-agent: *
Disallow: /_astro/

Sitemap: https://example.com/sitemap-index.xml

This only disallows crawling the compiled asset directory and declares the Sitemap URL; it does not block public posts. Ensure siteConfig.site and base are correct for custom domains and subpath deployments so absolute URLs point to the right location.

---

LLM Outputs and Privacy Boundaries

/llms.txt and /llms-full.txt are controlled by llmsConfig; see LLM-Friendly Content for section and extension settings. Both apply the same safety filters:

• Exclude draft: true posts;
• Exclude encrypted: true posts;
• Exclude posts matching excludeTags;
• Exclude posts matching excludeCategories.

/llms.txt is an index of titles, links, and descriptions. /llms-full.txt also combines public post bodies and processes <llm-only> and <llm-exclude> markers.

Disable the AI endpoints in src/config/llmsConfig.ts or the content repository's config/llms.yaml:

enable: false

Both URLs then return 404 and no static files are emitted. To keep only the index, set generateFull: false.

---

Pre-Release Checks

pnpm build

After building, inspect dist/ for rss.xml, atom.xml, sitemap-index.xml, robots.txt, and the configured llms.txt files. Confirm that:

1. siteConfig.site is the production domain, not a local preview URL;
2. base matches the hosting platform's path prefix;
3. RSS / Atom contain only sanitized encrypted entries, drafts are absent from production feeds, and both are excluded from LLM outputs;
4. Feed and Sitemap URLs return 200 and parse as valid XML.

Copyright

Copyright Ownership:matsuzaka-yuki

License under:Attribution 4.0 International (CC-BY-4.0)

Contributors: matsuzaka-yuki

Previous pageUmami Analytics

Next pageLLM-Friendly Content