ATLAS — AI Transport Logistics Agent Standard

The open-source MCP server that gives AI agents deep context about your logistics operations — without your data ever leaving your infrastructure.

---

The Problem

Enterprise logistics companies have years of operational data — emails, contracts, TMS records, carrier relationships, pricing history. AI agents need this context to be useful. But sharing raw data with external cloud services is a non-starter for compliance, legal, and security teams.

The result: AI stays shallow. Agents can’t negotiate from context. Every interaction starts from zero.

The Solution

ATLAS runs inside your security perimeter. It connects to your existing systems, indexes your data locally, and exposes a standardized MCP interface. Any AI agent can query ATLAS — getting deep operational context — without your data ever leaving your infrastructure.

[Your Company]                        [Cargofy / Any AI Agent]
  ├── Email                                      │
  ├── TMS                    MCP Protocol        │
  ├── ERP          ←─────────────────────────────┤
  ├── Contracts              (questions only,    │
  ├── Knowledge Base          no raw data out)   │
  └── ATLAS instance ────────────────────────────┘

Your data stays with you. Agents get the context they need.

---

Quick Start

Option 1: Docker (recommended)

docker run -p 3000:3000 cargofy/atlas

Open http://localhost:3000 — the Setup Wizard will guide you through initial configuration.

For production use with persistent data and AI features, see DOCKER.md.

Option 2: Claude Desktop

Add to your claude_desktop_config.json under mcpServers:

{
  "atlas": {
    "command": "docker",
    "args": ["run", "--rm", "-i", "cargofy/atlas", "node", "src/index.js"]
  }
}

Option 3: Run from source

git clone https://github.com/cargofy/ATLAS
cd ATLAS
npm install
cp config.example.yml config.yml   # edit with your settings
node seed.js                       # optional: load sample data
node src/ui-server.js              # Web UI + API on port 3000
# or
node src/index.js                  # MCP server on stdio

---

Web UI

ATLAS ships with a full web interface at http://localhost:3000: