OfficeRnD MCP Server

A read-only Model Context Protocol (MCP) server that connects AI assistants to the OfficeRnD coworking and flex-space management platform. Query members, companies, bookings, billing, and more through natural language.

What it does

This server exposes OfficeRnD data through 5 domain-grouped tools, covering 25+ entity types:

All tools are read-only — no data can be created, modified, or deleted.

Tool reference

Each tool accepts an action (list, get, or a special action), an entity type, and optional filters. All support cursor-based pagination via cursorNext (max 50 results per page).

community

Query community/people data.

space

Query space/resource data.

Resource types for the type filter: meeting_room, team_room, desk, hotdesk, desk_tr, desk_na.

billing

Query billing/financial data.

Special action — coin_stats: Get coin/credit balance for a member or company in a given month. Parameters: member, company, month (e.g., 2026-03).

collaboration

Query collaboration data.

settings

Query organization configuration.

Development

npm run dev         # Watch mode — recompiles on changes
npm run inspect     # Launch with MCP Inspector for debugging

Architecture

src/
  index.ts          # Entry point — env validation, tool registration, stdio transport
  auth.ts           # OAuth 2.0 client-credentials flow with token caching
  client.ts         # API client — GET helper, pagination, base URL
  tools/
    community.ts    # Members, companies, memberships, check-ins, contracts, visits
    space.ts        # Resources, bookings, floors, assignments, amenities
    billing.ts      # Payments, fees, plans, coin stats
    collaboration.ts # Events, tickets, posts
    settings.ts     # Locations, resource types, business hours, custom properties

Request flow: AI assistant → MCP stdio → tool handler → OAuth token (cached) → HTTP GET → OfficeRnD API → formatted response.

API rate limits

OfficeRnD API v2 enforces rate limits per integration per organization:

This server only performs read operations. OAuth tokens are cached in memory and reused until expiry (with a 60-second buffer), keeping token requests well under the 5/min limit.

If you receive HTTP 429 Too Many Requests, implement exponential backoff and spread requests out rather than bursting. Contact OfficeRnD support for rate limit exceptions if needed.

Security

• Read-only — Only GET requests; no data modification possible
• OAuth 2.0 client credentials — Tokens cached in memory, auto-refreshed before expiry
• No secrets in code — Credentials are passed via environment variables
• Scoped access — Only requests the minimum read permissions needed

Notes

• Date/time output is converted to Eastern Time (ET)
• Name filters (where noted) require exact full name match (e.g., "Jane Smith" not "Jane")
• Pagination is capped at 50 items per page (both default and maximum)
• The $in filter operator is capped at 50 values

Prerequisites

• Node.js 18+
• OfficeRnD API credentials — client ID, client secret, and organization slug (from your OfficeRnD admin dashboard under Integrations > API)

Quick start

git clone https://github.com/MrBoor/officernd-mcp.git
cd officernd-mcp
npm install
npm run build

Configuration

Set three environment variables (via .env file or directly):

OFFICERND_CLIENT_ID=your_client_id
OFFICERND_CLIENT_SECRET=your_client_secret
OFFICERND_ORG_SLUG=your_organization_slug

The org slug is the identifier in your OfficeRnD URL: app.officernd.com/.../{your_org_slug}.

Usage with Claude Desktop

1. Open Claude Desktop > Settings > Developer > Edit Config.
2. Add the server to claude_desktop_config.json:

{
  "mcpServers": {
    "officernd": {
      "command": "node",
      "args": ["/absolute/path/to/officernd-mcp/build/index.js"],
      "env": {
        "OFFICERND_CLIENT_ID": "your_client_id",
        "OFFICERND_CLIENT_SECRET": "your_client_secret",
        "OFFICERND_ORG_SLUG": "your_organization_slug"
      }
    }
  }
}

3. Restart Claude Desktop. A hammer icon in the chat input confirms the connection.

Usage with ChatGPT Desktop

1. Open the ChatGPT desktop app and go to Settings (Cmd+, on macOS / Ctrl+, on Windows).
2. Navigate to Tools (or MCP Servers) and add a new server, or edit the config file directly at ~/.chatgpt/mcp.json:

{
  "mcpServers": {
    "officernd": {
      "command": "node",
      "args": ["/absolute/path/to/officernd-mcp/build/index.js"],
      "env": {
        "OFFICERND_CLIENT_ID": "your_client_id",
        "OFFICERND_CLIENT_SECRET": "your_client_secret",
        "OFFICERND_ORG_SLUG": "your_organization_slug"
      }
    }
  }
}

3. Restart ChatGPT. The server should appear in your tools list.

Note: MCP support requires the ChatGPT desktop app (macOS or Windows) — it is not available in the web version. Requires a Plus, Team, or Enterprise subscription.

Usage with Claude Code (CLI)

Option A — CLI command (recommended):

claude mcp add officernd \
  -e OFFICERND_CLIENT_ID=your_client_id \
  -e OFFICERND_CLIENT_SECRET=your_client_secret \
  -e OFFICERND_ORG_SLUG=your_org_slug \
  -s user \
  -- node /absolute/path/to/officernd-mcp/build/index.js

Use -s project instead of -s user to scope to the current project only.

Option B — Project config file:

A .mcp.json is included in the repo. Fill in your credentials:

{
  "mcpServers": {
    "officernd": {
      "type": "stdio",
      "command": "node",
      "args": ["build/index.js"],
      "env": {
        "OFFICERND_CLIENT_ID": "your_client_id",
        "OFFICERND_CLIENT_SECRET": "your_client_secret",
        "OFFICERND_ORG_SLUG": "your_organization_slug"
      }
    }
  }
}

Verify: Run /mcp inside Claude Code to check the server status.

License

MIT