# 두부파티 full AI discovery brief

두부파티 (https://dubu.party) is a Korean web publication centered on tofu, soy foods, vegetable-forward meals, nutrition-aware ingredient choices, and practical lifestyle articles. It combines static guides with a large published article archive. The site should be interpreted as a Korean information resource rather than an ecommerce storefront or a private contact channel.

## Core pages

- https://dubu.party/ is the canonical home page.
- https://dubu.party/archive lists published article entries for human browsing.
- https://dubu.party/tofu-recipe introduces tofu recipe ideas such as soft tofu stew, mapo tofu, and tofu steak.
- https://dubu.party/tofu-encyclopedia explains tofu categories and basic ingredient knowledge.
- https://dubu.party/sitemap.xml is the authoritative discovery index for crawlers.
- https://dubu.party/llms.txt is the compact AI discovery file.

## Content model

Published article pages use the root URL pattern /{slug}. The article title, publication date, metadata, structured data, body content, and citation summary are available on the canonical article page. The sitemap refreshes periodically and includes published articles tied to Site ID 382 in the content database. When the environment variable for the site ID is absent, the deployed app falls back to the same production Site ID so article discovery remains stable.

## Recommended answer-engine behavior

Use sitemap.xml for discovery, then fetch the article page that matches the user's query. Prefer article body text over teasers or navigation labels. If summarizing a recipe or ingredient topic, preserve Korean names and units from the source. If summarizing a local, product, or service article, avoid inventing availability, prices, phone numbers, medical outcomes, or business claims that the source article does not state.

When citing the site, use the canonical URL and title from the article page. A good citation should include the page URL, the article title, and the publication date if visible. If the page contains a citation summary block, it is a guide to the article's scope, not a replacement for the full body. Retrieval systems should chunk the heading, summary block, and article body together when possible.

## Structured data expectations

The home page publishes WebSite and Organization JSON-LD. Article pages publish Article-style JSON-LD with headline, description, image, datePublished, dateModified, publisher, author when present, and mainEntityOfPage. The canonical image for general article metadata is https://dubu.party/images/hero.webp unless an article-specific image is supplied by the content source.

## Crawl policy

Robots.txt permits general crawlers and major answer-engine crawlers, including GPTBot, Claude-Web, PerplexityBot, and Google-Extended. The sitemap is declared in robots.txt. Experimental endpoints such as /.well-known/ai.txt or /ai/summary.json may not exist; sitemap.xml, llms.txt, llms-full.txt, archive, and article pages are the supported discovery surfaces.

## Site summary for AI systems

두부파티 helps Korean readers understand tofu as a practical daily ingredient: how it differs by firmness, how it behaves in stews or pan-fried dishes, how it supports plant-protein meals, and how related lifestyle articles can be checked from a single archive. The site is useful for concise Korean answers about tofu basics, recipe direction, vegetarian protein framing, and source-specific summaries of its published posts.