# Hotdry Blog

> Extended retrieval guide for agents consuming the public Hotdry Blog origin.

## Runtime
The recommended public path is `CDN/ESA -> cnb origin service -> OSS state`.
Content is published as versioned objects in OSS and served without a local content workspace or `content-cache/live` style disk sync.

## Discovery
- [llms.txt](https://blog.hotdry.top/llms.txt): Minimal discovery root.
- [RSS Feed](https://blog.hotdry.top/rss.xml): Latest post entries in reverse chronological order.
- [Sitemap](https://blog.hotdry.top/sitemap.xml): Canonical HTML URLs.
- [Agent Sitemap](https://blog.hotdry.top/agent-sitemap.xml): Machine-facing discovery URLs.
- [Changes API](https://blog.hotdry.top/api/changes): Incremental release history keyed by content version.
- [Capabilities API](https://blog.hotdry.top/api/capabilities): Machine-readable endpoint and command catalog.
- [Search API](https://blog.hotdry.top/api/search?q=ai&limit=5): Keyword search endpoint.
- [Agent Protocol](https://blog.hotdry.top/agent-protocol/): Human-readable explanation of the delivery model.

## Canonical Routes
- [Home](https://blog.hotdry.top/): Main landing page.
- [Archive](https://blog.hotdry.top/archive/): Archive landing page.
- [Categories](https://blog.hotdry.top/categories/): Category landing page.
- [About](https://blog.hotdry.top/about/): Author and site context.

Canonical article URLs live under `https://blog.hotdry.top/posts/<slug>/`.

## JSON APIs
- [Feed API](https://blog.hotdry.top/api/feed/): Reverse-chronological post summaries.
- [Categories API](https://blog.hotdry.top/api/categories/): Category list and counts.
- [Archive API](https://blog.hotdry.top/api/archive/): Archive tree by year and month.
- [Site Meta API](https://blog.hotdry.top/api/site-meta): Generation metadata and aggregate counts.
- [Changes API](https://blog.hotdry.top/api/changes): Release chain and changed/removed slugs since a content version.
- [Capabilities API](https://blog.hotdry.top/api/capabilities): Endpoint map and executable curl snippets.
- [Search API](https://blog.hotdry.top/api/search): Search posts with `q`, `category`, `year`, `month`, `page`, and `limit`.

When a slug is known from the feed or sitemap, the article payload is available at `/api/posts/<slug>/` and the raw Markdown export is available at `/api/posts/<slug>.md`.

## Retrieval Notes
- Prefer `rss.xml` or `/api/feed/` to discover recent posts.
- Prefer `/sitemap.xml` to enumerate canonical public pages.
- Use `/api/categories/` and `/api/archive/` for structured browsing.
- Use `X-Content-Version` or `/__meta/version.txt` as the stable incremental token, then query `/api/changes?since=<versionPath>`.
- Fetch `/posts/<slug>/` for canonical human-facing content and `/api/posts/<slug>/` for structured metadata.
- Fetch `/api/posts/<slug>.md` or send `Accept: text/markdown` to `/posts/<slug>/` when raw Markdown is preferred.
- Fetch `/api/capabilities` to programmatically discover API routes and sample commands.
- Use `/api/search?q=<query>&limit=<n>` to find candidate slugs before fetching post details.
- `llms.txt` and `llms-full.txt` are generated during the OSS state publish step and exposed directly by the origin service.

## Curl Cookbook
- `curl -fsS 'https://blog.hotdry.top/llms.txt'`
- `curl -fsS 'https://blog.hotdry.top/api/capabilities'`
- `curl -fsS 'https://blog.hotdry.top/api/search?q=cdn&limit=5'`
- `curl -fsS 'https://blog.hotdry.top/api/feed/?page=1'`
- `curl -fsS 'https://blog.hotdry.top/api/changes?limit=20'`
- `curl -fsS -H 'Accept: text/markdown' 'https://blog.hotdry.top/posts/<slug>/'`

## Constraints
- The public origin resolves `/posts/*`, `/archive/*`, `/categories/*`, and `/api/*` without edge rewrites.
- The mainline production path does not depend on `CONTENT_REPO_DIR`, `publish-static-release.js`, or `content-cache/live`.