Overview

The HorizonDataWave MCP Server is a lightweight Model Context Protocol (MCP) implementation for managing LinkedIn profiles and securely exposing user data to HorizonDataWave.ai. It implements an OAuth-style workflow and a small set of API endpoints so your application — or the HorizonDataWave.ai platform — can request permissioned access to a user’s LinkedIn profile, update profile data, and revoke access when needed.

This server is useful when you need a self-hosted bridge between LinkedIn data and the HorizonDataWave.ai account system. It centralizes consent management, token exchange, and profile synchronization, while keeping sensitive credentials and tokens in your controlled environment.

Features

OAuth-style authorization and token exchange for LinkedIn account access
Secure storage of access/refresh tokens (via environment-backed secret stores or local DB)
Endpoints to read LinkedIn profile data and update HorizonDataWave.ai account state
Webhook support for async events (token refresh, profile changes)
Simple configuration via environment variables and Docker
Lightweight codebase intended for self-hosting and integration into CI/CD pipelines

Installation / Configuration

Prerequisites: Node.js (14+), Docker (optional), a HorizonDataWave.ai API key and LinkedIn OAuth credentials.

Clone the repo:

git clone https://github.com/horizondatawave/hdw-mcp-server.git
cd hdw-mcp-server

Install dependencies and run locally:

npm install
npm run start

Example Docker run:

docker build -t hdw-mcp-server .
docker run -e PORT=3000 \
  -e HDW_API_KEY="your-hdw-api-key" \
  -e LINKEDIN_CLIENT_ID="li-client-id" \
  -e LINKEDIN_CLIENT_SECRET="li-client-secret" \
  -p 3000:3000 \
  hdw-mcp-server

Example docker-compose.yml:

version: '3.7'
services:
  mcp-server:
    image: hdw-mcp-server:latest
    ports:
      - "3000:3000"
    environment:
      - PORT=3000
      - HDW_API_KEY=${HDW_API_KEY}
      - LINKEDIN_CLIENT_ID=${LINKEDIN_CLIENT_ID}
      - LINKEDIN_CLIENT_SECRET=${LINKEDIN_CLIENT_SECRET}
      - SESSION_SECRET=${SESSION_SECRET}

Recommended environment variables:

PORT — server listen port (default 3000)
HDW_API_KEY — HorizonDataWave.ai API key for account updates
LINKEDIN_CLIENT_ID — LinkedIn OAuth client ID
LINKEDIN_CLIENT_SECRET — LinkedIn OAuth client secret
SESSION_SECRET — secret for encrypting session cookies or tokens
STORAGE_URL / DATABASE_URL — optional, for persistent token storage

Security notes: always supply secrets via environment variables or a secret manager. Run behind TLS in production.

Available Resources

API endpoints (common):

Method	Path	Description
GET	/mcp/authorize	Redirect users to LinkedIn consent screen (start auth flow)
GET	/mcp/callback	OAuth callback to exchange code for tokens
POST	/mcp/token	Exchange refresh token / issue MCP token to clients
GET	/mcp/profile	Fetch stored LinkedIn profile data for a user
POST	/mcp/update-profile	Push profile updates to HorizonDataWave.ai account
POST	/mcp/disconnect	Revoke tokens and disconnect account
POST	/mcp/webhook	Optional webhook receiver for background events

Typical responses use JSON and standard HTTP status codes. Authentication for MCP consumer apps can be via short-lived JWTs issued by the server after successful OAuth exchange.

Useful links:

GitHub: https://github.com/horizondatawave/hdw-mcp-server
HorizonDataWave.ai docs: (use your account documentation link)

Use Cases

Onboard a user’s LinkedIn profile to HorizonDataWave.ai
- Flow: Your app redirects user to GET /mcp/authorize. After consent, the MCP server exchanges the authorization code and persists tokens. The server fetches profile data and calls /mcp/update-profile to sync the user’s HorizonDataWave.ai account.
Regular profile synchronization
- Use a scheduled job to call GET /mcp/profile for each user and compare fields. When changes are detected, POST /mcp/update-profile to push updates into HorizonDataWave.ai so models have up-to-date context.
Token refresh and revocation
- The server handles refresh tokens automatically. If a user disconnects, POST /mcp/disconnect will revoke LinkedIn tokens and remove stored data, then notify HorizonDataWave.ai to remove the associated context.
Programmatic access for integrations
- Third-party tools can request temporary MCP tokens from POST /mcp/token (after appropriate authorization) to read a user’s profile without storing long-lived LinkedIn credentials.

Example curl: start auth (client-side pattern)

# Redirect the browser to this URL to start consent flow
https://your-mcp-host.example.com/mcp/authorize?redirect_uri=https://your-app.example.com/after-auth

Example curl: fetch profile (server-to-server, after obtaining a client token)

curl -H "Authorization: Bearer <mcp-token>" \
  https://your-mcp-host.example.com/mcp/profile

Notes for Developers

Treat this server as an integration layer: keep HorizonDataWave.ai logic separate from LinkedIn API handling where possible.
Implement robust error handling for API rate limits and token expiry.
Use HTTPS and rotate secrets regularly.
If you intend to scale, plug a database (Postgres, Redis) for token/session persistence and configure webhooks for event-driven updates.

HorizonDataWave MCP Server: LinkedIn Profile Management