Getting Started

Get up and running with OneMCP in minutes. The CLI provides the fastest way to start chatting with AI models connected to your APIs.

Prerequisites

Java 21 (if you are running natively)
Docker Desktop or equivalent (if you are running within Container)
Depending on the chosen model, you will also need an API Key.
- OpenAI (Requires an API Key)
- Anthropic (Requires an API Key)
- Gemini (Requires an API Key)
- Ollama

Step 1: Install the CLI

Install the Gentoro OneMCP CLI with a single command:


curl -sSL https://raw.githubusercontent.com/Gentoro-OneMCP/onemcp/main/packages/cli/install.sh | bash

The installer will download and set up everything you need.

Step 2: Start Chatting

Launch the interactive chat interface:


onemcp chat

On your first run, the setup wizard will guide you through:

Choosing your AI provider (OpenAI, Anthropic, or Gemini)
Entering your API key
Selecting a starting point (built-in ACME Analytics example or your own API)

You’ll see an interactive interface like this:


╔══════════════════════════════════════╗
║     Gentoro OneMCP - Chat Mode       ║
╚══════════════════════════════════════╝

Handbook: ecommerce-api
Provider: openai
MCP URL: http://localhost:8080/mcp
Type 'exit' to quit, 'clear' to clear history, 'switch' to change handbook
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

💡 Mock Server Active - Try These Example Queries:

  > Show me electronics sales in California last quarter.
  > List top customers by revenue.
  > Compare revenue trends by region.

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

? You:

Step 3: Try Example Queries

With the built-in ACME Analytics mock service, try queries like:

“Show me electronics sales in California last quarter”
“List top customers by revenue”
“What are the top-selling products this month?”

The AI will automatically call the appropriate APIs and return structured results.

What Just Happened?

The CLI automatically:

Started the OneMCP server (port 8080)
Launched the TypeScript runtime (port 7070)
Started the ACME Analytics mock server (port 8082)
Connected your chosen AI provider
Indexed the handbook documentation and API specs