MCP: Let Your AI Publish for You

27 March 2026#mcp#ai#claude#cursor

The Model Context Protocol (MCP) is an open standard that lets AI applications interact with external tools. scsiwyg ships a dedicated MCP server that gives AI IDEs direct access to your blog.

Instead of copying curl commands into your terminal, you can just tell your AI: "Publish a post about today's standup." And it does.

What the MCP provides

The scsiwyg MCP server exposes tools for blog management and newsletters — everything from publishing posts to sending newsletters, managing contributors, and generating images.

Tool	What it does
`get_my_site`	Returns your site info (username, title, bio)
`list_posts`	Lists all your published posts
`get_post`	Gets a single post by slug
`publish_post`	Creates and publishes a new post
`update_post`	Updates an existing post
`delete_post`	Deletes a post
`update_site`	Updates your site title, bio, or theme
`generate_image`	Generate a cover image via DALL-E 3
`upload_image`	Upload an image to Vercel Blob storage

Every tool uses your Bearer token for authentication. The MCP server is stateless — each request is independent.

Setting up with Claude Code

Add the MCP server to your Claude Code config:

bash

claude mcp add scsiwyg \
  --transport http \
  --url https://scsiwyg.com/api/mcp \
  --header "Authorization: Bearer YOUR_TOKEN_HERE"

Or add it to .claude/settings.json:

json

{
  "mcpServers": {
    "scsiwyg": {
      "type": "http",
      "url": "https://scsiwyg.com/api/mcp",
      "headers": {
        "Authorization": "Bearer YOUR_TOKEN_HERE"
      }
    }
  }
}

Setting up with Claude Desktop

In your Claude Desktop config file:

json

{
  "mcpServers": {
    "scsiwyg": {
      "command": "npx",
      "args": [
        "mcp-remote",
        "https://scsiwyg.com/api/mcp",
        "--header",
        "Authorization: Bearer YOUR_TOKEN_HERE"
      ]
    }
  }
}

What it looks like in practice

Once the MCP is configured, conversations like this just work:

You: Write a short post about the new auth system I just shipped. Tag it with "engineering" and "auth".

Claude: I'll publish that for you. [calls publish_post with slug "new-auth-system", title "Shipping the New Auth System", body with markdown, tags ["engineering", "auth"]] Published! Your post is live at /yourname/new-auth-system.

You: Actually, change the title to "Auth System v2 is Live"

Claude: [calls update_post with the new title] Updated.

No browser. No copy-paste. No context switching. The AI goes directly from your intent to a published post.

Why MCP instead of just the API?

The API is great for scripts and automation. But MCP is about conversation. When your AI IDE has MCP access, it can:

Discover your blog's current state (what's published, what's in draft)
Draft a post based on your conversation context
Publish with a natural language confirmation
Iterate by updating the post based on your feedback

The API is a tool. MCP turns that tool into a capability your AI assistant has permanently.

𝕏 Post