# Hotdry Blog LLM Integration Guide

## Scope
- This site serves both human HTML and agent-oriented Markdown.
- Research pages contain an explicit `Agent Context` block.
- The phrase “latest deep research” is only asserted when freshness checks pass.

## Transport Negotiation
- For any page URL, request Markdown with one of:
  - `Accept: text/markdown`
  - `x-convert-to-markdown: true`
- Optional headers:
  - `x-respond-with: markdown|html`
  - `x-content-options: strip_links,strip_images`

## Converter Endpoint
- Endpoint: https://blog.hotdry.top/cdn-cgi/content-converter/markdown
- Supported source inputs:
  - query `source` or `url` (site-relative or same-origin absolute URL)
  - header `x-source-url` (same-origin only)
- External URL fetching is blocked by default in this deployment.

## Discovery
- llms.txt: https://blog.hotdry.top/llms.txt
- agent sitemap: https://blog.hotdry.top/agent-sitemap.xml
- canonical sitemap: https://blog.hotdry.top/sitemap.xml

## Reliability Notes
- JSON APIs remain unchanged for backward compatibility.
- Markdown routes use ETag + Cache-Control for CDN friendliness.
- Agent snapshots are pre-generated into local content cache.