Charming MCP - Generate & Host Interactive Web Apps via AI Agents

Charming is a hosted MCP server that generates, hosts, and updates interactive web apps. Connect it to any MCP-compatible AI client and ask it to build an app - you get back a live URL in seconds. OAuth with Dynamic Client Registration is automatic - clients that support it show a one-time consent screen.

Think of it as "Vercel for agents" - your AI can go from idea to deployed interactive web app without touching a CLI, managing hosting, or configuring DNS.

What It Does

• App generation: AI agents describe an app and Charming builds + deploys it
• Live hosting: Every generated app gets a public URL immediately
• Iterative updates: Ask the agent to modify the app - Charming redeploys
• OAuth-native: No API keys to manage - Dynamic Client Registration handles auth
• Interactive apps: Full-stack web apps, not static pages

Key Tools

Tool	Description
charming_create_app	Generate and deploy a new interactive web app
charming_update_app	Modify and redeploy an existing app
charming_list_apps	List all apps created by this agent
charming_delete_app	Remove an app
charming_get_app_url	Get the live URL for an app

Quick Start

1. Add to MCP Client

{
  "mcpServers": {
    "charming": {
      "url": "https://api.charming.mcp.so/mcp"
    }
  }
}

2. First App

Once connected, just ask your agent:

"Build me a calculator app with a dark theme" "Create a markdown editor with live preview" "Make a poll/voting app for my team"

The agent calls charming_create_app, and Charming returns a live URL.

3. Iterate

"Add a history panel to the calculator" "Change the theme to match our brand colors"

The agent calls charming_update_app with the existing app ID.

Hermes Agent Integration

# ~/.hermes/config.yaml
mcp_servers:
  charming:
    url: "https://api.charming.mcp.so/mcp"

First connection triggers OAuth consent in your browser - one-time only.

Use Cases for Business Operators

1. Rapid prototyping: Describe a dashboard, CRM view, or data tool and get a working version in seconds
2. Internal tools: Build utility apps for your team without writing code - "make a PTO request form"
3. Client demos: Generate interactive demos during sales calls - "build a demo of what the analytics would look like"
4. Landing pages: Spin up event pages, waitlist forms, or micro-sites on demand
5. A/B test variants: Generate multiple page variants and iterate on winners

Integration with CorpusIQ

CorpusIQ is the connector layer that lets AI assistants read the systems a business already runs on - accounting, payments, CRM, email, files and analytics - and answer with the source named. It is read-only by design: it retrieves data and never writes to, or acts in, a connected system, which is what makes it safe to point at live accounts.

This page is part of the CorpusIQ directory, the reference we maintain for the MCP server and agent-skill ecosystem. Run alongside CorpusIQ, a third-party server like this one can be combined with a company's own business data in a single answer.

See what CorpusIQ connects

Limitations

• No custom domains (yet): Apps are served from Charming's domain
• No backend persistence: Apps are frontend-focused; persistent data requires external services
• Early stage: 9 GitHub stars as of July 2026 - feature set expanding
• Hosted only: No self-hosted option currently - apps live on Charming's infrastructure

Example: Building a Dashboard

User: "Build me a sales dashboard with a bar chart showing monthly revenue"

Agent → charming_create_app({
  name: "Sales Dashboard",
  description: "Monthly revenue bar chart with filters",
  requirements: "Bar chart, dark theme, responsive, month filter"
})

Charming → { app_id: "abc123", url: "https://abc123.charming.app" }

User: "Add a line chart for customer growth and make the background white"

Agent → charming_update_app({
  app_id: "abc123",
  changes: "Add line chart overlay for customer growth, switch to light theme"
})

Charming → { url: "https://abc123.charming.app" }  // Updated live

Category

Development - Charming bridges the gap between AI agent capabilities and deployed, interactive web applications. It's a "deploy target" for agents, similar to how Vercel is a deploy target for developers.

---

Discovered via mcp.so on July 24, 2026. Verified listing. GitHub: tambo-labs/charming-mcp. 9 stars.