@scaffold-hbar-ui/mcp-server

Model Context Protocol (MCP) server that exposes scaffold-hbar-ui documentation, TypeScript props, hook signatures, and example app code so AI agents can answer with accurate imports and APIs.

Features

• Tools: search_components, search_hooks, get_component_props, get_hook_signature, get_component_example, get_hook_example, get_installation_guide, get_theming_guide
• Resources: scaffold-hbar-ui:// URIs for MDX docs, examples, and JSON API listings
• Prompts: scaffold-hbar-ui/use-component, scaffold-hbar-ui/hedera-dapp-setup
• Transports: stdio (default) for Cursor / Claude Desktop, or Streamable HTTP with --http for remote agents

Build

From the monorepo root:

pnpm install
cd packages/mcp-server && pnpm build

This runs extract (generates src/generated/metadata.json from docs and sources) and compiles TypeScript.

Quick test (MCP Inspector)

The fastest way to explore the server is the official MCP Inspector — a visual debugger that launches the server and lets you click through tools, resources, and prompts:

# from packages/mcp-server, after `pnpm build`
npx @modelcontextprotocol/inspector node ./dist/esm/bin/scaffold-hbar-ui-mcp.js

Open the printed http://localhost:6274 URL, then try:

• Tools → search_components (query hedera), get_component_props (name HbarInput), get_hook_signature (name useBalance)
• Resources → scaffold-hbar-ui://api/components, or the templated scaffold-hbar-ui://components/Address
• Prompts → scaffold-hbar-ui/use-component

Use it in an MCP client (Cursor, Claude Desktop, …)

The server speaks MCP over stdio, so any MCP-capable client can spawn it. Add an entry to the client's MCP config, then fully restart the client. See the official guide on connecting to local MCP servers for the general flow.

After publishing to npm (no local checkout needed):

{
  "mcpServers": {
    "scaffold-hbar-ui": {
      "command": "npx",
      "args": ["-y", "@scaffold-hbar-ui/mcp-server"]
    }
  }
}

For local development, point at the built binary instead (run pnpm build first):

{
  "mcpServers": {
    "scaffold-hbar-ui": {
      "command": "node",
      "args": ["/absolute/path/to/scaffold-hbar-ui/packages/mcp-server/dist/esm/bin/scaffold-hbar-ui-mcp.js"]
    }
  }
}

Where to put the config:

• Cursor — ~/.cursor/mcp.json (global) or .cursor/mcp.json inside a project. Working in this repo? Copy .cursor/mcp.json.example to .cursor/mcp.json — it uses ${workspaceFolder}, so no path edits are needed.
• Claude Desktop — Settings → Developer → Edit Config, which opens ~/Library/Application Support/Claude/claude_desktop_config.json (macOS) or %APPDATA%\Claude\claude_desktop_config.json (Windows).

Restart the client, confirm the scaffold-hbar-ui server shows as connected, then ask something like "Use the scaffold-hbar-ui tools to show the HbarInput props and a usage example." If it doesn't connect, check the client's MCP logs (Claude Desktop: ~/Library/Logs/Claude/mcp*.log).

HTTP (remote)

PORT=3100 node ./dist/esm/bin/scaffold-hbar-ui-mcp.js --http

Optional auth:

SCAFFOLD_HBAR_UI_MCP_KEY=secret node ./dist/esm/bin/scaffold-hbar-ui-mcp.js --http

Clients must send Authorization: Bearer <secret>.

Connect with:

{
  "mcpServers": {
    "scaffold-hbar-ui": {
      "url": "http://localhost:3100/mcp"
    }
  }
}

Streamable HTTP clients must send Accept: application/json, text/event-stream (MCP requirement). Example:

curl -X POST http://localhost:3100/mcp \
  -H "Content-Type: application/json" \
  -H "Accept: application/json, text/event-stream" \
  -d '{"jsonrpc":"2.0","id":1,"method":"initialize","params":{"protocolVersion":"2024-11-05","capabilities":{},"clientInfo":{"name":"curl","version":"1"}}}'

Docker

See Dockerfile in this package. Build from the repository root so the extract step can read docs/ and packages/:

docker build -f packages/mcp-server/Dockerfile -t scaffold-hbar-ui-mcp .