MCP Jibun Server

English | 简体中文

This is a Model Context Protocol (MCP) server that allows AI Agents to read posts from Jibun instances. It is also compatible with Ech0 instances.

Features

Provides the following tools for AI Agents to call:

• get_jibun_posts: Get the latest posts from a Jibun (or Ech0) instance. Supports specifying source (source), count (count), and page number (page).

• list_jibun_sources: List all configured Jibun (or Ech0) instance sources.

Related MCP server: riocloud-reader

Development and Deployment

1. Local Development and Cloudflare Workers

We use wrangler for local development and Cloudflare deployment.

Configuration: Configure environment variables in the vars field of the wrangler.jsonc file:

// wrangler.jsonc
{
  "vars": {
    "JIBUN_INSTANCES": "[{\"name\":\"...\",\"url\":\"...\"}]"
  }
}

Common Commands:

• Start local server:

  pnpm dev

• Deploy to Cloudflare:

  pnpm run deploy

2. Deploy to Vercel / Netlify

This project is based on the Hono framework and can automatically adapt to Vercel and Netlify runtime environments. It can be deployed with one click by connecting to a Git repository, but please note to configure the correct environment variable in the management panel: JIBUN_INSTANCES.

Format: A JSON string containing a list of Jibun instances.

Example:

[
  {"name": "E.m. じぶん", "url": "https://jibun.elecmonkey.com"},
  {"name": "River's Lighthouse", "url": "https://river177.com"}
]

Client Configuration

MCP Server URL:

• Cloudflare: https://<your-worker-subdomain>.workers.dev

• Vercel: https://<your-project>.vercel.app

• Netlify: https://<your-site>.netlify.app

Supports / or /mcp paths.