AI Agents & MCP¶
Run Open Quantum jobs from AI coding agents using the Model Context Protocol (MCP).
The openquantum-mcp package is a local stdio MCP server on top of the Core SDK. Compatible clients (Claude, Cursor, VS Code, Grok, and others) can discover backends, submit circuits, wait for completion, and summarize results using your SDK key — without hand-writing boilerplate for each iteration.
What you get¶
- Discover organizations, backends, and credit balances
- Prepare a job and inspect the credit quote before spend
- Submit OpenQASM circuits to real QPUs and simulators
- Wait / poll for status (queues can be long; the agent re-polls)
- Summarize measurement counts and fetch calibration on demand
There is a single MCP server for all clients. Client apps differ only in how you register the process and environment variables.
How it works¶
AI client (Claude, Cursor, …)
│ MCP (stdio)
▼
openquantum-mcp (local process on your machine)
│
▼
openquantum-sdk → Open Quantum APIs
Credentials stay on your machine in the MCP process environment.
Prerequisites¶
- Python 3.10+
- An Open Quantum account and SDK key
- An MCP-capable agent or IDE
Next steps¶
- Installation — install from PyPI and verify
- Client setup — wire Claude, Cursor, and other hosts
- Tools & spend policy — tool reference and credit safeguards