Carrierscore MCP server
{
"mcpServers": {
"carrierscore": {
"type": "http",
"url": "https://mcp.carrierscore.io/mcp"
}
}
}{
"mcpServers": {
"carrierscore": {
"type": "http",
"url": "https://mcp.carrierscore.io/mcp"
}
}
}{
"mcpServers": {
"carrierscore": {
"type": "http",
"url": "https://mcp.carrierscore.io/mcp"
}
}
}{
"servers": {
"carrierscore": {
"type": "http",
"url": "https://mcp.carrierscore.io/mcp"
}
}
}{
"mcpServers": {
"carrierscore": {
"type": "http",
"url": "https://mcp.carrierscore.io/mcp"
}
}
}file .mcp.json (in your project root)file claude_desktop_config.jsonfile ~/.cursor/mcp.jsonfile .vscode/mcp.jsonfile ~/.codeium/windsurf/mcp_config.jsonversion 0.1.0
What it does
CarrierScore exposes FMCSA motor-carrier risk scores and timestamped evidence reports built from public safety data. It is for freight brokers and AI booking agents who need to assess carrier risk and maintain compliance documentation at the moment of carrier selection.
- Look up a carrier's identity and FMCSA safety rating by US DOT number to verify a rate confirmation.
- Generate a timestamped evidence report (Montgomery file) documenting carrier risk at the point of booking for compliance retention.
- Screen a batch of up to 100 carriers by DOT number to identify risk scores and active flags before load assignment.
Tools this server exposes
| Tool | Does |
|---|---|
| carrier_lookup | Returns carrier identity, legal name, DBA, operating status, FMCSA safety rating, fleet size, address, and registration dates by US DOT number. |
| carrier_score | Returns a risk score (0–100, higher = riskier) with component breakdown, hard flags such as active out-of-service orders or missing insurance, and data-sufficie |
| montgomery_file | Returns a timestamped carrier-selection evidence report for compliance documentation; text format on free tier, JSON on paid tiers. |
| monitor_carriers | Screens up to 100 DOT numbers in a single call, returning batch score summary and active flags for each carrier. |
Before you enable it
An MCP server runs with whatever access you give it. Start with a development account rather than production, and check which tools the server exposes before you let an agent call them unattended.