🤖 Gins Blog MCP Usage Guide

This blog implements the Model Context Protocol (MCP), allowing AI agents (like Claude Desktop, OpenClaw, or Cursor) to directly interact with your blog's content and data.

🚀 Capabilities

Once connected, your AI agent can:

• Search Posts: Find articles by keywords (search_posts).
• Draft Content: Create new post drafts directly in your database (draft_post).
• Analyze Comments: Fetch recent comments for moderation or sentiment analysis (analyze_comments).
• Check Analytics: View improved traffic stats (check_analytics).

---

🔌 Connection Setup

Option 1: Claude Desktop

To use this with the Claude Desktop App:

1. Locate your Claude config file:

   • macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
   • Windows: %APPDATA%\Claude\claude_desktop_config.json

2. Add the following configuration (replace /absolute/path/to/... with your actual path):

   {
     "mcpServers": {
       "gins-blog": {
         "command": "node",
         "args": ["/Users/your-username/path/to/gins-blog/scripts/mcp-client.js"],
         "env": {
           "BLOG_URL": "https://your-deployed-blog.com"
         }
       }
     }
   }
   

3. Restart Claude Desktop. You should see a 🔌 icon indicating the tool is connected.

---

Option 2: OpenClaw / Cursor

For agents that support openclaw.json or standard MCP over stdio:

1. Ensure you have openclaw.json in your project root.

2. Edit openclaw.json to point BLOG_URL to your production URL:

   {
     "mcpServers": {
       "gins-blog": {
         "command": "node",
         "args": ["scripts/mcp-client.js"],
         "env": {
           "BLOG_URL": "https://your-deployed-blog.com" 
         }
       }
     }
   }
   

3. Instruct your agent: "Load the MCP server defined in openclaw.json".

---

✅ How to Verify Locally

Before connecting to an external agent, you can verify your MCP server is working correctly using the provided test script.

1. Start your local server:

   npm run dev
   

2. Run the test script in a new terminal:

   node scripts/test-mcp.js
   

   You should see output similar to:

   🔌 Connecting to MCP Endpoint: http://localhost:4321/api/mcp...
   1️⃣  Testing 'initialize' (List Tools)...
   ✅ Tools Found: search_posts, draft_post, analyze_comments, check_analytics
   ...
   

---

🗣️ Example Prompts

Here are some things you can ask your connected AI:

Content Creation

"Check my recent posts using search_posts. Based on topics I haven't covered lately, draft a new blog post title and outline about 'The Future of Edge Computing'."

"Write a draft post about my trip to Kyoto using the draft_post tool. Title it 'Kyoto Diaries'."

Analytics & Management

"Use check_analytics to tell me how my blog is performing today. Which post is the most popular?"

"Analyze the last 10 comments with analyze_comments. Are there any spam or negative comments I should be worried about?"

---

🛠️ Troubleshooting

• Connection Refused: Ensure BLOG_URL in your config points to the deployed version of your blog (e.g., https://blog.ichimarugin728.com), not localhost, unless you are running npm run dev locally.
• Method Not Found: Ensure you have redeployed your blog after adding the MCP endpoint code.