Skip to content
UniqueCapital

Machine Interface / UniqueResearch

For AI

Machine entry points and a suggested read order. Chinese is the primary archive. People can keep reading in the English research library, the data and rankings list, or the events and awards list. Models should follow the order, then fetch one Markdown article.

Suggested order

  1. /for-aiStart on this page to see which URLs exist and how to fetch one Markdown article.
  2. /llms.txtRead the short guide and the latest three data points.
  3. /llms-full.txtRead this when you need bounded context. It is not the full archive.
  4. /for-modelsRead the citation rules before quoting.
  5. /research/index.jsonChoose an article by title, date, and tags, then use its markdownUrl.
  6. /research/latest.mdRead the rolling summary for recent updates, or use rss.xml.

Machine URLs

Public machine entry points. Article bodies are not in these index files.
URLPurpose
/for-aiDirectory of machine URLs and the suggested read order.
/llms.txtShort guide: who we are, read steps, three latest data points, and product links.
/llms-full.txtBounded context: definitions, citation rules, latest points, and archive pointers. No article bodies.
/for-modelsCitation rules: keep the UniqueResearch byline, original date, and source link. Do not infer an individual author.
/research/index.jsonChinese article metadata: title, date, byline, source, markdownUrl, and content gaps. No bodies.
/research/latest.mdRecent titles, dates, on-site links, and a one-line summary.
/research/rss.xmlRSS 2.0 for the same rolling window as latest.md.
/sitemap.xmlPublic page list, including article HTML, interface pages, and llms-full.txt.
/robots.txtCrawl rules. Content-Signal allows search, ai-input, and ai-train.

Fetch one Markdown article

The HTML page and the Markdown file are generated from the same archived text. The Markdown response is text/markdown; charset=utf-8 with YAML frontmatter. Images and video have no transcript.

  • Canonical path: /research/{slug}/markdown, English /en/research/{slug}/markdown. markdownUrl in index.json points here.
  • Alias: /research/{slug}.md and /en/research/{slug}.md return the same text. /research/latest.md stays a rolling summary.
  • Article HTML includes <link rel="alternate" type="text/markdown"> and the HTTP header Link: <https://ffcap.cn/research/20261023a0eccc00/markdown>; rel="alternate"; type="text/markdown".
  • The article URL does not switch on Accept: text/markdown. Browsers always receive HTML.

One current example (2026-10-24): https://ffcap.cn/research/20261023a0eccc00/markdown or https://ffcap.cn/research/20261023a0eccc00.md

Copyable prompt

Agent prompt
You are an assistant reading UniqueResearch public material. Start with https://ffcap.cn/for-ai and https://ffcap.cn/llms.txt. For a short context, also read https://ffcap.cn/llms-full.txt (it does not include every article). Then fetch https://ffcap.cn/research/index.json for titles, dates, and Markdown URLs. Read one article at https://ffcap.cn/research/{slug}/markdown or https://ffcap.cn/research/{slug}.md. English articles are listed at https://ffcap.cn/en/research. When quoting, follow https://ffcap.cn/for-models: keep the publication name UniqueResearch (非凡产研), the original publication date, and the source link. Do not infer an individual author.
curl llms.txt
curl -sL https://ffcap.cn/llms.txt
curl latest.md
curl -sL https://ffcap.cn/research/latest.md
curl one Markdown article
curl -sL https://ffcap.cn/research/{slug}.md

Citation rules are on the citation page. This page does not grant extra content rights.