Pendientes — Todo Manager MCP App

What it does

Renders a full interactive todo list UI directly inside your MCP-capable AI client. Tasks are created, edited, completed, and deleted through a clean Vanilla TypeScript interface — no external app or browser tab needed. All data persists locally in todos.json, including creation timestamps and completion timestamps.

Key features

• Full CRUD in-client — add, edit, delete, and complete tasks without leaving your AI assistant
• Filter views — switch between All / Pending / Completed task lists
• Timestamp tracking — records exact creation time and completion time for every task
• Persistent storage — todos survive server restarts via todos.json
• Dual transport — runs as HTTP server (port 3001) or stdio for different MCP clients

Installation

Claude Desktop (claude_desktop_config.json):

{
  "mcpServers": {
    "Pendientes": {
      "command": "npx",
      "args": ["pendientes-mcp-app", "--stdio"]
    }
  }
}

HTTP mode (Claude.ai web or any HTTP-capable client):

npx pendientes-mcp-app
# Server available at http://localhost:3001/mcp

Add http://localhost:3001/mcp in Claude.ai Settings → Integrations.

VS Code Copilot (settings.json):

{
  "github.copilot.chat.mcp.servers": {
    "pendientes": {
      "command": "npx",
      "args": ["pendientes-mcp-app", "--stdio"]
    }
  }
}

Supported hosts

Works with any MCP client supporting stdio or Streamable HTTP transport.