quantum-mcp 0.1.0

Quantum Computing and Multi-Agent Orchestration MCP Server for Azure Quantum

Quantum MCP Server

MCP server providing quantum computing capabilities to Claude Code via the Model Context Protocol. Integrates Azure Quantum (gate-based) and D-Wave (quantum annealing) backends with multi-agent orchestration.

Project Status

Phase	Focus	Status	Tests
Phase 0	Quantum Foundation (VQE, QAOA, Kernels, Q#)	Complete	~174
Phase 1	Multi-Agent Orchestration	Complete	~180
Phase 3	Quantum Orchestration (QUBO, D-Wave)	In Progress	~88

Total: 442 tests passing

Architecture

Azure Quantum (Gate-Based)     D-Wave (Annealing)
   VQE, QAOA, Kernels, Q#         QUBO Solving
            │                          │
            └──────────┬───────────────┘
                       │
         ┌─────────────┴───────────────┐
         │   Multi-Agent Orchestration │
         │   Routing | Consensus       │
         │   Decomposition             │
         └─────────────┬───────────────┘
                       │
         ┌─────────────┴───────────────┐
         │   Backend Abstraction       │
         │   Azure | D-Wave | Exact    │
         └─────────────┬───────────────┘
                       │
                ┌──────┴──────┐
                │  MCP Server │
                └──────┬──────┘
                       │
                ┌──────┴──────┐
                │ Claude Code │
                └─────────────┘

Requirements

• Python 3.11+
• uv package manager
• Azure Quantum workspace (optional, for cloud execution)
• D-Wave Leap account (optional, for quantum annealing)

Installation

git clone https://github.com/ryanmat/quantum_mcp.git
cd quantum_mcp
uv sync --all-extras

For D-Wave support:

uv sync --extra dwave

Configuration

Copy .env.example to .env:

cp .env.example .env

Azure Quantum (Gate-Based)

AZURE_QUANTUM_WORKSPACE_ID=your-workspace-id
AZURE_QUANTUM_RESOURCE_GROUP=your-resource-group
AZURE_QUANTUM_SUBSCRIPTION_ID=your-subscription-id

# Service Principal (optional)
AZURE_TENANT_ID=your-tenant-id
AZURE_CLIENT_ID=your-client-id
AZURE_CLIENT_SECRET=your-client-secret

D-Wave Quantum Annealing

DWAVE_API_TOKEN=DEV-xxx...
DWAVE_SOLVER=Advantage_system6.4

Execution Settings

DEFAULT_BACKEND=ionq.simulator
MAX_SHOTS=1000
BUDGET_LIMIT_USD=10.0
LOG_LEVEL=INFO

Running the Server

uv run python -m quantum_mcp

MCP Tools

Connectivity

Tool	Description
ping	Test server connectivity

Backend Management

Tool	Description
quantum_list_backends	List Azure Quantum backends
quantum_estimate_cost	Estimate job cost before submission

Circuit Simulation

Tool	Description
quantum_simulate	Local Qiskit simulation (bell, ghz, qft, custom)

Quantum Algorithms

Tool	Description
quantum_vqe	Variational Quantum Eigensolver for ground state energy
quantum_qaoa	QAOA for combinatorial optimization (MaxCut)
quantum_kernel	Quantum kernel matrix for ML classification
quantum_run_qsharp	Execute Q# quantum programs
quantum_anneal	Solve QUBO via D-Wave quantum annealing

Tool Examples

VQE (Ground State Energy)

quantum_vqe(
    num_qubits=2,
    ansatz_type="ry",
    hamiltonian_type="h2",
    optimizer="COBYLA",
    shots=1000
)

QAOA (Combinatorial Optimization)

quantum_qaoa(
    edges=[[0, 1], [1, 2], [0, 2]],
    layers=2,
    optimizer="COBYLA",
    shots=1000
)

Quantum Annealing (QUBO)

quantum_anneal(
    qubo={"0,0": -1, "1,1": -1, "0,1": 2},
    num_reads=100,
    backend_type="auto"
)

Q# Execution

quantum_run_qsharp(
    code='''
    {
        use q = Qubit();
        H(q);
        let result = M(q);
        Reset(q);
        result
    }
    ''',
    shots=100
)

Multi-Agent Orchestration

Agent Types

• ClaudeAgent - Anthropic Claude
• OpenAIAgent - OpenAI GPT
• LocalAgent - Ollama local models
• ToolAgent - Tool-based wrapper

Routing Strategies

• CapabilityRouter - Routes by agent capabilities
• LoadBalancingRouter - Distributes load
• LearnedRouter - ML-based with performance tracking
• QUBORouter - QUBO formulation for routing
• QAOARouter - Quantum-enhanced via QAOA
• DWaveRouter - D-Wave annealing for routing

Consensus Mechanisms

• VotingConsensus - Majority voting
• WeightedMergeConsensus - Weighted merging
• DebateConsensus - Multi-round debate

Project Structure

quantum_mcp/
├── src/quantum_mcp/
│   ├── __main__.py              # Entry point
│   ├── config.py                # Configuration
│   ├── server.py                # MCP server
│   ├── agents/                  # Agent implementations
│   ├── backends/                # Quantum backends
│   │   ├── annealing.py         # Annealing protocol
│   │   ├── dwave.py             # D-Wave Leap
│   │   └── exact_solver.py      # Classical fallback
│   ├── circuits/                # Quantum algorithms
│   │   ├── vqe.py               # VQE
│   │   ├── qaoa.py              # QAOA
│   │   ├── kernels.py           # QML kernels
│   │   └── qsharp.py            # Q#
│   ├── client/                  # Azure Quantum client
│   ├── orchestration/           # Multi-agent system
│   │   ├── router.py            # Routing strategies
│   │   ├── consensus.py         # Consensus mechanisms
│   │   ├── decomposer.py        # Task decomposition
│   │   ├── qubo.py              # QUBO formulation
│   │   ├── qaoa_router.py       # QAOA router
│   │   └── dwave_router.py      # D-Wave router
│   └── tools/                   # MCP tools
│       ├── registration.py      # Tool registration
│       └── algorithm_tools.py   # Algorithm tools
├── tests/
│   ├── unit/                    # Unit tests (442)
│   └── integration/             # Integration tests
├── docs/                        # Internal documentation
├── mcp.json                     # MCP configuration
├── pyproject.toml               # Project config
└── .env.example                 # Environment template

Testing

# Unit tests
uv run pytest tests/unit -v

# With coverage
uv run pytest tests/unit --cov=quantum_mcp

# Integration tests (requires credentials)
uv run pytest tests/integration -v

Cost Management

Default budget: $10.00 USD

Backend costs:

• IonQ Simulator: Free
• IonQ Hardware: ~$0.01/shot
• D-Wave: Per-second QPU time
• Quantinuum: Varies by plan

Use quantum_estimate_cost before Azure hardware submissions.

MCP Configuration for Claude Code

Add to your ~/.claude/mcp.json:

{
  "mcpServers": {
    "quantum": {
      "command": "uv",
      "args": ["--directory", "/path/to/quantum_mcp", "run", "python", "-m", "quantum_mcp"],
      "env": {
        "DWAVE_API_TOKEN": "DEV-xxx..."
      }
    }
  }
}

Dependencies

Core:

• azure-quantum, azure-identity
• qiskit, qiskit-aer
• qsharp
• mcp
• pydantic

Optional:

• anthropic, openai (agents)
• dwave-ocean-sdk, dimod (annealing)

Dev:

• pytest, pytest-asyncio, pytest-cov
• ruff, mypy

Related Projects

predictive-insights - Temporal X-DEC for metric prediction

• BiGRU-XVAE-DEC with temporal attention
• 5 semantic operational state clusters

Future integration planned at Phase 14 (MoE).

License

MIT