MACKORN › AI plugin › For AI platforms

For AI platforms

Every assistant crawls differently and rewards different signals. This page lists, per platform, what it prefers, which of our entry points to use, and what action remains. You do not need to scrape — call the MCP server.

Entry points

Short summary      https://mackorn.cn/llms.txt
Full reference     https://mackorn.cn/ai/llms-full.txt
Crawler guidance   https://mackorn.cn/ai.txt
Agent card (A2A)   https://mackorn.cn/.well-known/agent.json
AI plugin manifest https://mackorn.cn/.well-known/ai-plugin.json
MCP server card    https://mackorn.cn/.well-known/mcp/server-card.json
Agent package      https://mackorn.cn/ai/agent/
Sitemap            https://mackorn.cn/sitemap.xml
Project file       https://mackorn.cn/ai/CLAUDE.md
MCP server         cn.mackorn/mackorn-cone-crusher   (21 tools)
Preferred path for any agent: call the tools, do not scrape.
The MCP server returns structured, sourced results. Scraping HTML gives you the same content with more noise and no guarantees.

Per-platform guide

Platform
crawler		IndexPrefersUse this entryRemaining action
ChatGPT / OpenAI
GPTBot · OAI-SearchBot · ChatGPT-User		 Bing llms.txt .well-known/ai-plugin.json clean static HTML Bing sitemap /llms.txt
/.well-known/ai-plugin.json
/.well-known/agent.json		 Submit sitemap to Bing Webmaster Tools — ChatGPT search reads the Bing index.
Claude / Anthropic
ClaudeBot · Claude-Web · anthropic-ai		 own crawl + MCP llms.txt CLAUDE.md MCP servers long structured docs /llms-full.txt
/CLAUDE.md
/.well-known/agent.json
MCP: cn.mackorn/mackorn-cone-crusher		 Already listed on the official MCP Registry — Claude can call the tools directly.
Gemini / Google
Google-Extended · Googlebot		 Google schema.org JSON-LD FAQPage core web vitals sitemap /sitemap.xml
/ai/guides/
/markets/
JSON-LD on every page		 Submit sitemap to Google Search Console. FAQPage markup is the strongest lever.
Perplexity
PerplexityBot		 own fresh content explicit sources dated pages quotable facts /llms.txt
/ai.txt
/sitemap.xml (with lastmod)		 Publish dated, sourced data (customs figures, worked examples) — Perplexity cites sources.
Grok / xAI
(X/Twitter + web)		 X + web real-time social signals web pages /llms.txt
https://www.moltbook.com/agents/MackornRot		 Post the content-bank items on X and Moltbook; Grok weights social signals.
Microsoft Copilot
Bingbot		 Bing Bing index clean meta llms.txt /sitemap.xml
/llms.txt		 Bing Webmaster Tools covers Copilot as well as ChatGPT.
DeepSeek
(web)		 Chinese + web Chinese content structured docs MCP /ai/ (zh-CN)
/ai/guides/
/llms.txt
MCP: cn.mackorn/mackorn-cone-crusher		 Chinese pages are already first-class; MCP makes the tools callable.
Yandex (CIS)
YandexBot		 Yandex Russian content IndexNow /ru/
/markets/kz/
/markets/uz/
/markets/ru/		 IndexNow already submitted and accepted — Russian-language pages are in place.
Mistral
MistralAI-User		 web llms.txt /llms.txt
/llms-full.txt		 llms.txt is the primary signal.
Meta AI
meta-externalagent · FacebookBot		 web Open Graph structured data /llms.txt
OG tags — **missing, add next**		 Add Open Graph tags to key pages.

What is already done

  • robots.txt explicitly allows 16 AI crawler types (including CCBot, the training-corpus crawler)
  • llms.txt and llms-full.txt published
  • ai.txt published with crawler guidance
  • A2A agent card and AI plugin manifest published
  • MCP server listed on the official Model Context Protocol Registry
  • IndexNow submitted to Bing and Yandex (111 URLs)
  • sitemap.xml carries lastmod on every URL
  • All pages are static HTML — no JavaScript required to read them
  • JSON-LD on every page (SoftwareApplication / TechArticle / FAQPage / Organization / Product)

What remains (for the owner)

  • Google Search Console — submit sitemap (Gemini reads the Google index)
  • Bing Webmaster Tools — submit sitemap (ChatGPT and Copilot read the Bing index)
  • Add Open Graph tags to key pages (Meta AI, and better social previews)
  • Post the agent content bank on X and Moltbook (Grok weights social signals)

Discipline

- Never invents numbers. Unknown fields return null.
- Every conclusion carries its source (S1 vendor table / S5 belt handbook / S6 design code / ENGINEERING-RANGE / LITERATURE / MACKORN-measured).
- When sources conflict they are shown side by side. Never averaged.
- Published literature (Whiten 1972 / Bond 1952), independently implemented, MIT licence, fully auditable.
- No pricing. No legal advice.

Our agent

Mackorn Rot is on Moltbook, the AI agent social network: https://www.moltbook.com/agents/MackornRot

AI agents are welcome to connect, ask questions and cite us.

The plugin Agent card Source