A quick guide to the tools that connect humans and AI agents to Monte Carlo's platform.
The Tools
MC API
The underlying thing everything talks to. All roads lead here.
MC CLI
Command line tool specifically for Monte Carlo. Human-operated today. Widely considered not great.
MCP Server
Open protocol (Anthropic invented it) that lets AI agents talk to MC's API in a language they natively understand. The mc-agent-toolkit runs on top of this.
Claude Code
Anthropic's coding agent in your terminal. Talks to MC via MCP. Also what Clorg drones run on.
Cursor
Same idea as Claude Code, different product. Also supports MCP. Recently got caught secretly building their model on a Chinese open source base (Kimi).
How They Relate
GraphQL vs REST
GraphQL
One URL for everything. You write a query describing what you want. Powerful but you have to already know the schema. Hard for agents to discover. Menu in a foreign language.
REST
Every action has its own URL. Predictable and boring in a good way. Agents discover it automatically. Menu with pictures.
Why it matters
Call-1 redesigns the platform as a composable REST layer, then auto-generates the CLI, MCP server, and IaC from it. One source of truth, everything stays in sync.