Screen any vessel by IMO for sanctions, ownership opacity and vetting grade. Keyless check included.
An MCP server for the ArcNautical API: screen any commercial vessel by IMO number for sanctions, ownership opacity and a vetting grade — from Claude, ChatGPT, Cursor, VS Code, Windsurf or any MCP client.
Two tools need no API key at all. Add it, ask "is IMO 9274446 sanctioned?", get the answer.
Two ways to use it: a remote endpoint at https://mcp.arcnautical.com/mcp (nothing to install — claude.ai, ChatGPT, Claude Code, Cursor) or a local server over stdio with npx -y @arcnautical/mcp (Claude Desktop and every other stdio client).
IMO 9274446 HS STAR: sanctions RED — 4 confirmed matches on vessel identifier.
Ownership opacity MEDIUM. Vetting grade E (unacceptable). Checked 2026-09-12T08:10:40Z.
https://mcp.arcnautical.com/mcp — MCP Streamable HTTP, stateless, JSON responses. No authentication is required for check_vessel and find_port; the keyed tools take your API key as an Authorization: Bearer header on the MCP request (an X-API-Key or arcnautical-api-key header is accepted too, for gateways that name the header themselves — never a query parameter).
claude.ai — Settings → Connectors → Add custom connector → URL https://mcp.arcnautical.com/mcp, no authentication. The two keyless tools are available at once.
ChatGPT — Settings → Connectors → Create (developer mode) → server URL https://mcp.arcnautical.com/mcp, no authentication.
Claude Code
claude mcp add --transport http arcnautical https://mcp.arcnautical.com/mcp
claude mcp add --transport http arcnautical https://mcp.arcnautical.com/mcp --header "Authorization: Bearer arc_live_…"
Cursor — .cursor/mcp.json:
{
"mcpServers": {
"arcnautical": {
"url": "https://mcp.arcnautical.com/mcp",
"headers": { "Authorization": "Bearer arc_live_…" }
}
}
}
Anything with curl — the endpoint is plain JSON-RPC over POST:
curl -s https://mcp.arcnautical.com/mcp -H 'Content-Type: application/json' \
-d '{"jsonrpc":"2.0","id":1,"method":"tools/call","params":{"name":"check_vessel","arguments":{"imo":"9274446"}}}'
The remote endpoint is a thin client of the same public API: a keyless call counts against your 100-per-hour allowance (the caller's address is forwarded), and a keyed call is metered on your key exactly as a direct API call would be. GET /mcp answers 405 — there is no server-to-client stream to hold — and GET https://mcp.arcnautical.com/health reports the endpoint's version.
Requires Node 18+. No install step — every client below runs it with npx.
Claude Desktop — claude_desktop_config.json:
{
"mcpServers": {
"arcnautical": {
"command": "npx",
"args": ["-y", "@arcnautical/mcp"]
}
}
}
Claude Code
claude mcp add arcnautical -- npx -y @arcnautical/mcp
Cursor / Windsurf / VS Code — same command / args shape in the client's MCP settings.
To unlock the full record, batches and voyage scoring, add an API key to the env block:
{
"mcpServers": {
"arcnautical": {
"command": "npx",
"args": ["-y", "@arcnautical/mcp"],
"env": { "ARCNAUTICAL_API_KEY": "arc_live_…" }
}
}
}
Keys are self-serve at https://arcnautical.com/arcnautical.html#/developer-api — no approval step. A key includes 5,000 vessel screenings and 10,000 voyage assessments a month.
| Tool | Key? | What it does |
|---|---|---|
check_vessel | no | Sanctions status (OFAC SDN, EU, UN, UK OFSI, OpenSanctions), ownership-opacity score, A–E vetting grade for one IMO. 100/hour per IP. |
find_port | no | Resolve a port name or country to UN/LOCODEs — call before score_voyage. |
screen_vessel | yes | The full screening record: every match with its source list, programme and confidence class, the graded vetting factors, per-source freshness, a retained record id. Same hull asked again today replays free. |
screen_vessels | yes | Batch up to 50 IMOs and wait for the results. |
get_screening | yes | Retrieve a stored record by id (retained ten years). |
score_voyage | yes | Route risk between two LOCODEs: score, level, drivers, confidence, missing sources. |
get_usage | yes | Remaining allowance and limits for the configured key. |
coverage_complete: false on a GREEN means one supplementary list was unavailable: re-screen before relying on it.assessed: false means ownership and vetting are defaults, not findings.screen_vessel covers that case.checkedAt / screened_at time with any answer.Every resource-creating call sends an Idempotency-Key derived from the question — the hull and the UTC day — not from the attempt. Asking about the same vessel twice in a day returns the stored record and spends nothing. The API's docs measured 42% of one integration's monthly screenings as same-day repeats billed only because their client generated a random key per call; this server does not do that.
| Variable | Default | Purpose |
|---|---|---|
ARCNAUTICAL_API_KEY | unset | Enables the keyed tools (stdio; the remote endpoint reads the Authorization header instead). |
ARCNAUTICAL_BASE_URL | https://arcnautical.com | Override for testing. |
PORT / HOST | 3005 / all interfaces | arcnautical-mcp-http only. |
MCP_TRUST_PROXY | unset | arcnautical-mcp-http only: 1 reads the caller's address from X-Real-IP / CF-Connecting-IP set by a reverse proxy. Only when nothing but the proxy can reach the port. |
npx -y -p @arcnautical/mcp arcnautical-mcp-http serves POST /mcp on port 3005 — useful when your assistants run in a network that cannot spawn local processes but can reach one internal URL. The Dockerfile in this repository builds the same thing; it is what mcp.arcnautical.com runs.
com.arcnautical/mcp) · Smithery · npmnpm install
npm test # builds, then drives dist/cli.js over stdio AND createHttpServer() over Streamable HTTP
# (a local stub API asserts the forwarded caller address and Bearer key; one call hits prod keyless)
npm run start:http # the remote endpoint on http://localhost:3005/mcp
MIT © ArcNautical
Source-derived launch command. Check the maintainer’s required arguments and credentials before running:
npx -y @arcnautical/mcpMerge this template into ~/Library/Application Support/Claude/claude_desktop_config.json. Keep existing servers. Add any arguments, credentials, and permissions required by the maintainer; this template has not been install-tested.
{
"mcpServers": {
"com-arcnautical-mcp": {
"command": "npx",
"args": [
"-y",
"@arcnautical/mcp"
]
}
}
}Restart Claude Desktop completely for changes to take effect. Confirm the server appears connected in the client’s tool list, then try a read-only example from its documentation.
Claude Desktop setup referencecom.arcnautical/mcp works with any MCP-compatible client. Copy the config snippet from the Configuration section above and add it to the file shown for your client, then restart the application.
~/Library/Application Support/Claude/claude_desktop_config.jsonRestart Claude Desktop completely for changes to take effect.~/.cursor/mcp.jsonRestart Cursor for changes to take effect..vscode/mcp.jsonReload VS Code window for changes to take effect.~/.codeium/windsurf/mcp_config.jsonRestart Windsurf for changes to take effect..mcp.jsonSave at the project root, then start Claude Code in that project and review the MCP server approval prompt. Keep real credentials out of shared files.