Grounding LLMs in Maritime Data: Using MCP for Port Intelligence

An Intelligence Briefing for the Port of Rotterdam, from a Single Prompt

The Model Context Protocol (MCP) enables LLMs to call external APIs during conversations to retrieve live data. In a test case for the Port of Rotterdam, this system autonomously executed tool calls across seven distinct API endpoints to synthesize a briefing.

Why This Matters

Standard LLMs rely on static training data, leading to hallucinations that can render technical reports useless if IMO numbers or inspection records are fabricated. Grounding models via MCP shifts the AI’s role from reproducing memory to orchestrating live database queries, reducing trust issues by linking outputs directly to primary sources like AIS position data and EU MRV emissions reports.

Key Insights

• The model processed 20 recent vessel movements at Rotterdam between 18:11–19:59 UTC, identifying four seagoing vessels including the STENA FORETELLER and THULELAND.
• EU MRV framework provides mandatory monitoring of CO2; for example, the THULELAND showed a 43% YoY CO2 reduction in 2024.
• Paris MoU New Inspection Regime (NIR) can trigger detailed inspections based on history, as seen with STENA FORETELLER receiving 7 deficiencies in April 2024 and 6 more in January 2025.
• The vesselapi-mcp server allows any compatible AI client (Claude Code, Cursor) to access maritime registry and safety data via Node.js 18+.

Working Examples

.mcp.json configuration file for adding VesselAPI as an MCP server.

{
"mcpServers": {
"vesselapi": {
"command": "npx",
"args": ["-y", "vesselapi-mcp"],
"env": {
"VESSELAPI_API_KEY": "YOUR_API_KEY"
}
}
}
}

Practical Applications

• ): Ship brokers use consolidated profiles (particulars, emissions, PSC history) for due diligence; manually querying multiple systems is a common inefficiency.
• ): Commodity trading desks track vessel movements to estimate cargo flows; relying on outdated static snapshots leads to inaccurate supply disruption forecasts.

References:

• https://dev.to/vessel_api/an-intelligence-briefing-for-the-port-of-rotterdam-from-a-single-prompt-2a7
• vesselapi.com