MCP server exposing the Ramoira brand schema specification as tools and resources.
Any MCP-compatible client (Claude Desktop, Cursor, Windsurf, etc.) can connect to this server and use it to generate, validate, or reason about Ramoira brand schemas — without needing the CLI installed.
| Tool | Description |
|---|---|
get_spec_overview |
High-level explanation of the schema format, its five layers, and key concepts. Start here. |
get_layer_spec(layer) |
Full field definitions for one layer: identity, narrative, voice, commercial, or governance. |
get_valid_enums() |
All valid enum values for every constrained field, with field paths. |
get_example_schema() |
A high-quality worked example (Rolex) showing what well-populated fields look like. |
search_spec(query) |
Keyword search across all spec documentation. |
| URI | Description |
|---|---|
ramoira://spec |
Spec overview as a plain text resource. |
ramoira://enums |
Valid enum values as a JSON resource. |
ramoira/brand-schema-spec ← the spec (source of truth)
│
▼
ramoira/mcp ← this package
serves the spec as MCP tools so any LLM can consume it
│
├─► Claude Desktop / Cursor / any MCP client
│ Any LLM session can look up the spec on demand
│
└─► ramoira/cli (internally)
ramoira init uses the same tool set locally
to give Claude spec access during schema generation
The CLI's generator mirrors this tool set locally (no network dependency) so ramoira init works offline. This server is for external LLM workflows — any agent that needs to generate or validate a schema can connect here instead of inlining the spec in its prompt.
npm install -g @ramoira/mcpOr run directly without installing:
npx @ramoira/mcpAdd to ~/Library/Application Support/Claude/claude_desktop_config.json (macOS) or %APPDATA%\Claude\claude_desktop_config.json (Windows):
{
"mcpServers": {
"ramoira": {
"command": "npx",
"args": ["@ramoira/mcp"]
}
}
}Add to your MCP settings:
{
"ramoira": {
"command": "npx",
"args": ["@ramoira/mcp"]
}
}The server currently uses stdio transport, which works for local clients. To serve it publicly over HTTP, replace the transport in src/server.ts:
import { StreamableHTTPServerTransport } from "@modelcontextprotocol/sdk/server/streamableHttp.js";
const transport = new StreamableHTTPServerTransport({ port: 3000 });
await server.connect(transport);Then deploy to any Node.js host. Clients connect to https://mcp.ramoira.com (or wherever it's hosted) via the MCP HTTP client.
npm install
npm run dev # run via tsx (no build step)
npm run build # compile to dist/