Introduction

For years, developers have automated everything—deployments, tests, builds, CI/CD… But strangely, blog publishing still feels stuck in 2012.

If you write a blog today, you still need to:

• Write it


• Format it


• Copy/paste it to every platform


• Fix spacing issues


• Add tags


• Publish manually


• Repeat for each platform

As someone who writes often, I kept asking myself:

"Why can't I write a blog once… and publish it everywhere?"

So I built exactly that.

And yes—this blog post was published through the tool itself.

🎯 Introducing: BlogCaster MCP

BlogCaster MCP is a Model Context Protocol (MCP) server that lets you publish your blog directly from an AI agent like Claude or Cursor.

Write your blog in AI. Edit your blog in AI. Refine it in AI. Then run:

publishPost(

  title: "...",

  contentMarkdown: "...",

  platforms: ["hashnode", "devto"]

)

Done. Your post goes live instantly.

No dashboards. No back-and-forth. No formatting hell.

✔️ What's Completed Right Now (Phase 1)

1. Multi-platform publishing

Supported today:

• Hashnode


• Dev.to

The system is built to scale, so new platforms plug in easily.

2. AI-native writing workflow

You can:

• Write a blog with AI


• Edit paragraphs


• Ask for improvements


• Generate examples


• Fix tone/SEO/style


• And when ready → publish

Everything stays inside Claude or Cursor.

3. Markdown → Publish pipeline

Give it any .md file (README, notes, drafts) and the server publishes it seamlessly.

🧱 Architecture Overview

Here's how everything fits together:

User (Claude/Cursor)

       ↓

BlogCaster MCP Worker

       ├─→ Durable Objects (Secure Token Storage)

       ├─→ Markdown Processor

       ├─→ Platform Adapters

       │   ├─→ Hashnode API

       │   └─→ Dev.to API

       └─→ Response Handler

BlogCaster handles:

• Authentication


• Markdown formatting


• API calls


• Error handling


• Platform-specific quirks

AI tools only need to call the MCP method.

🚧 Work in Progress (Phase 2)

This is where things get exciting.

I'm now building full blog management, not just publishing.

1. Get All Posts (List All Posts)

View all posts from multiple platforms inside your AI editor:

• Titles


• Drafts


• Publish status


• IDs


• Links

Useful for selecting posts to update or remove.

2. Update Any Post

Ask the AI:

"Rewrite my old recursion article in a friendlier tone and update it everywhere."

The tool will:

• Pull your existing post


• Let you edit or ask AI to rewrite


• Push the updated version back to every selected platform

3. Delete Posts

Clean up old drafts or wrong uploads with one command:

deletePost(platform: "hashnode", id: "...")

4. More Platforms Coming

Next on the list:

• Medium


• Ghost


• WordPress


• Hashnode publications


• Blogger API


• Dev.to organizations

5. AI-Powered Enhancements

• Auto-generate tags


• Auto-generate cover images


• SEO optimization


• Fix headings, formatting, table of contents


• "Show diff before updating"

🌐 Why MCP Changes Everything

I could have built this as a CLI or API.

But with MCP:

• Claude/Cursor treat BlogCaster like a native tool


• AI can access your blog platforms securely


• No more context switching


• The entire writing → editing → publishing workflow happens in one place

MCP turns your AI into a blogging command center.

🎉 How This Blog Was Published

To make this blog fully meta:

1. I wrote and refined it inside Claude


2. I connected BlogCaster MCP through mcp-remote
   


3. I saved my API tokens


4. I ran:
   

publishPost(

  title: "I Built an MCP Server That Publishes Blogs Automatically",

  contentMarkdown: "<entire post>",

  platforms: ["hashnode", "devto"]

)

This post appeared instantly on my blogs.

That's It

Phase 1: Done ✅

Phase 2: Coming soon 🚀

Connect:

Building in the open. Come say hi.