Quantum Suitability Validator MCP

MCP server that screens quantum computing POC proposals against expert decision rules -- before your agent escalates any initiative to a committee, allocates budget, or routes to a specialist.

What it does

Enterprise innovation agents and R&D workflow agents process backlogs of proposed technology initiatives tagged as potential quantum computing candidates. Before escalating any candidate to a human committee, allocating POC budget, or routing to a quantum specialist, the agent calls quantum_assess_problem to produce an auditable triage verdict.

This server is refusal-first by design. It downgrades or refuses more often than it approves. Every verdict is auditable and machine-readable.

Tools

quantum_assess_problem (Free: 5/month, no key required)

Screens a quantum computing proposal using an expert-validated four-dimensional scoring framework. Returns:

• verdict: SCIENTIFICALLY_RECOMMENDED_NOW | COMMERCIALLY_RECOMMENDED_NOW | INVESTIGATE_FURTHER | PREMATURE | NOT_QUANTUM_AMENABLE
• four_scores: scientific_fit (40% weight), hardware_feasibility (25%), advantage_potential (25%), commercial_relevance (10%), composite -- four independent 0.0-1.0 scores so a scientifically valid investigation is never confused with proven commercial advantage
• advantage_claim_level: NONE | HYPOTHESISED | EXPERIMENTAL_SIGNAL | BENCHMARK_SUPPORTED | PRODUCTION_VALIDATED
• suitability_score: 0.0-1.0 (equal to four_scores.composite)
• confidence_score: 0.0-1.0
• problem_class: combinatorial_optimisation | portfolio_optimisation | molecular_simulation | ml_kernel | cryptography_pqc | sampling_monte_carlo | other
• dominant_blockers: specific reasons why the problem fails screening
• hype_flags: detected hype language patterns
• baseline_question: always "What is your classical baseline today, and what metric must improve for this to matter?"
• next_best_action: specific actionable recommendation
• agent_action: ESCALATE_TO_POC | ROUTE_TO_SIMULATOR | DEFINE_BASELINE_FIRST | REJECT | REQUEST_MORE_INFORMATION

quantum_readiness_report (Pro only)

Full auditable Quantum Readiness Report, weighted by audience profile (RESEARCH, ENTERPRISE, or INVESTOR -- the same problem legitimately scores differently by profile). Everything from quantum_assess_problem plus:

• recommended_workflow: CLASSICAL_ONLY | HYBRID | SIMULATOR_ONLY | ANNEALING_PATH | GATE_MODEL_VARIATIONAL | INSUFFICIENT_INFORMATION
• formulation_guidance: QUBO/Ising/variational suitability, estimated binary variables, penalty dominance risk
• hardware_recommendations: hardware family fit scores with access routes (D-Wave Leap, IBM Cloud, IonQ Cloud)
• error_budget_assessment: viability against current noise floors
• classical_baseline_assessment: baseline strength and minimum benchmark requirement
• validation_plan: ordered steps for technical review board submission
• refusal_reason: populated when the report declines to recommend a path forward
• commercial_reality_statement: populated for ENTERPRISE and INVESTOR profiles -- states plainly that production advantage over classical has not yet been broadly demonstrated

Connect

HTTP (Railway -- no install)

{"type": "http", "url": "https://quantum-suitability-validator-mcp-production.up.railway.app"}

stdio (npm -- requires ANTHROPIC_API_KEY)

npx quantum-suitability-validator-mcp

Harness Integration

Note: this server exposes tools at /mcp not the root URL.

Claude Code / Claude Desktop (.mcp.json)

{
  "mcpServers": {
    "quantum-suitability-validator": {
      "type": "http",
      "url": "https://quantum-suitability-validator-mcp-production.up.railway.app/mcp"
    }
  }
}

LangChain (Python)

from langchain_mcp_adapters.client import MultiServerMCPClient
client = MultiServerMCPClient({
    "quantum-suitability-validator": {
        "url": "https://quantum-suitability-validator-mcp-production.up.railway.app/mcp",
        "transport": "http"
    }
})
tools = await client.get_tools()

OpenAI Agents SDK (Python)

from agents import Agent, HostedMCPTool
agent = Agent(
    name="Assistant",
    tools=[HostedMCPTool(tool_config={
        "type": "mcp",
        "server_label": "quantum-suitability-validator",
        "server_url": "https://quantum-suitability-validator-mcp-production.up.railway.app/mcp",
        "require_approval": "never"
    })]
)

LangGraph

Same as LangChain above — langchain-mcp-adapters works with LangGraph natively.

Pricing

• Free: 5 quantum_assess_problem calls/month per IP -- no API key required
• Pro: $199/month -- unlimited quantum_assess_problem + full quantum_readiness_report
• Enterprise: $499/month -- volume + SLA

Upgrade: kordagencies.com

Legal

AI-assisted triage -- NOT a substitute for experimental physicist review. Results are for informational and planning purposes only and do not constitute expert quantum computing advice. Full terms: kordagencies.com/terms.html

Kord Agencies Pte Ltd, Singapore