elessar

If your website is your own code, you do not need a CMS connection. Everything a brand has published to its site on elessar is readable through a public, read-only content API, keyed by the brand's site handle. There is no key and no login, because published content is public. A small MCP server wraps the same API so a coding agent can pull the articles into your site, sources and all.

The handle

The handle is the blog address chosen under Settings → Your website. For a blog at tryelessar.com/b/acme, the handle is acme. Only articles someone has pressed "Publish to your site" on are visible; drafts and approved-but-unpublished articles are not.

The content API

All addresses start with https://tryelessar.com/api/v1/site/ followed by the handle.

The site: the brand's name and how many articles are published.
The articles: /articles, newest first, with limit and a before cursor for the next page.
One article: /articles/<slug>, with the title, the body as Markdown and as HTML, the meta title and description, the images with their alt text, the sources, and the citations that tie each claim to a passage in a source.
An image: /articles/<slug>/images/<slot>.
A feed: /feed.xml, for feed readers and anything that takes RSS.

Everything is JSON except the feed and the images. Read it at build time, at request time, or on a schedule; it is cached for a few minutes at our end, so an article published just now appears shortly after.

The MCP server

The package is called @tryelessar/mcp. It runs on your machine, needs Node 18 or later, and exposes four tools: list the articles, get one article (Markdown, HTML or both, with images, sources and citations), get the sources, and the site's details. It cannot write anything.

For Claude Code, the one command to run is:

claude mcp add elessar -- npx -y @tryelessar/mcp

For Cursor or any client that reads an mcpServers block, add an entry that runs npx -y @tryelessar/mcp. Then ask your agent to add the latest article from your handle to your blog, with its sources.

Keep the sources

Every article elessar publishes cites its sources, and each claim resolves to a passage in one of them. The API returns the sources and the citations with the body. Render the sources with the article, so your readers can check what it says. That is the point of the product, and it is what makes the article worth publishing.

If it goes wrong

A 404 for the site means the handle is wrong or none has been chosen yet.
An empty article list means nothing has been published to the site. Approving is not publishing; press "Publish to your site" on the article.
The MCP server reads tryelessar.com by default. Set ELESSAR_API_URL only if you were told to.