The Curator · Product

Agent Integrations (MCP)

The Curator publishes a Model Context Protocol server. Connect it once and your AI assistant — ChatGPT, Claude, Cursor, or anything else that speaks MCP — can read our articles, blog posts, FAQs and press releases directly, with citations, instead of guessing from a stale crawl.

Server endpoint

https://lisaiceland.com/mcp

Transport: Streamable HTTP. Authentication: OAuth 2.1 (your client registers itself — there is no API key to copy). Access is read-only.

What MCP is, briefly

The Model Context Protocol is an open standard that lets an AI assistant call tools on a remote server the same way a person clicks a link. Instead of pasting our pages into a chat window, you add The Curator as a connector once; the assistant then queries the live library whenever it needs a fact from it.

How to connect

  1. Open your assistant's connector settings — in ChatGPT and Claude this is under Settings → Connectors; in Cursor and most developer tools it is an mcpServers entry.
  2. Add a new remote (HTTP) MCP server and paste https://lisaiceland.com/mcp.
  3. Your client will open a sign-in page. Sign in with your The Curator account (or create one — it is free), then approve the connection on the consent screen. Nothing is shared until you approve.
  4. Ask the assistant to list the available tools. If you see list_brands, you are connected.

Config-file clients accept the equivalent JSON:

{
  "mcpServers": {
    "curator": {
      "url": "https://lisaiceland.com/mcp"
    }
  }
}

Tools you get

  • list_brandsLists the three publications in the network with their domains and editorial focus. Start here — the ids it returns are the `brand` argument for every other tool.
  • search_articlesFull-text search across the knowledge library. Returns titles, slugs, summaries and hub categories.
  • get_articleFetches one article in full, including every section, so your assistant can quote or summarise it accurately.
  • list_blog_postsLists editorial blog posts for a brand, newest first, with author and publication date.
  • get_blog_postFetches the complete body of a single blog post by slug.
  • list_faqsReturns published FAQ entries — useful for support questions about the site itself.
  • list_press_releasesReturns published press releases and announcements.

Things worth trying

  • "Search The Curator for everything on Innovation Discovery and summarise the current thinking."
  • "Pull the latest The Curator blog post and turn it into a five-bullet brief for my team."
  • "Using The Curator articles only, build a reading path for someone new to this topic."
  • "What does The Curator say in its FAQs about how the content is produced?"

Access, privacy and limits

  • Read-only. No tool can create, edit or delete anything. Your assistant cannot post comments, change your account or spend credits.
  • You approve every client. Connections use OAuth 2.1 with an explicit consent screen, and requests run under your own account permissions — never someone else's.
  • Published content only. Drafts, moderation queues and other members' private data are out of scope.
  • Attribution. Content stays ours; quoting is welcome with a link back to the source page. See our Terms and Editorial Policy.
  • Fair use. The endpoint is rate-limited. For bulk or commercial use, get in touch first.

Troubleshooting

  • Client says "unauthorized". The sign-in either was not completed or expired — remove the connector and add it again.
  • No tools listed. Make sure the URL ends in /mcp and that the connector is a remote HTTP server, not a local command.
  • Empty results. Pass a valid brand id — run list_brands first.
  • Still stuck? Visit the support centre.

Related