# 28-Jul-2026 — improve AI content discovery per site CMS [ref #55](https://github.com/nbarrett/ngx-ramblers/issues/55)

## [build 799](https://github.com/nbarrett/ngx-ramblers/actions/runs/30315657879) — [commit 0e6b566](https://github.com/nbarrett/ngx-ramblers/commit/0e6b566b9f512be02b6f95acd8367094843dad89)

_____

## What's new

Public CMS pages were already available as HTML, Markdown and JSON, but the best starting points were hard for tools to find. Each site now advertises a clearer path for assistants and crawlers: a richer `llms.txt`, an AI landing page, and a release feed when the site actually publishes release notes. Discovery only promotes release hubs that exist in that site's CMS - no invented short URLs or staging-only paths assumed for every group.

## At a glance

- Every site still builds `/llms.txt` dynamically, with documentation hubs and concrete export examples
- New `/for-ai` guide for assistants (Markdown by default; also HTML and JSON)
- Release notes and documentation hubs appear in AI discovery only when those public CMS pages exist on the site
- New `/api/public/releases` feed (also at `/api/releases.json`) lists recent releases with Markdown/JSON/HTML links when the site publishes them; returns 404 when it does not
- Sites advertise `llms.txt` and `/for-ai` via link tags and HTTP headers
- Hard-coded `/release-notes` and `/release-notes-for-humans` aliases removed
- Sitemap includes real discovery entry points for that site