Security Orchestra

50+ specialized AI agents & 8 compound chains for data center critical power infrastructure.

---

Installation

Claude Desktop (hosted)

Add to your claude_desktop_config.json:

{
  "mcpServers": {
    "security-orchestra": {
      "url": "https://security-orchestra-orchestrator.onrender.com",
      "transport": "sse"
    }
  }
}

Claude Code (CLI)

claude mcp add security-orchestra https://security-orchestra-orchestrator.onrender.com --transport sse

npx (Smithery)

npx -y @smithery/cli@latest mcp add robotfleet-hq/security-orchestra

Install via Smithery.

Restart your MCP client. All tools are immediately available.

---

What It Does

Security Orchestra provides deterministic, standards-based calculations for every phase of data center infrastructure — from site selection to commissioning. Every agent returns structured JSON with citations to applicable codes (NEC, NFPA, IEEE, ASHRAE, Uptime Institute).

Tool Categories

Compound Chains (8)

Chains run multiple agents sequentially and return a combined result:

---

Supported Protocols

Security Orchestra exposes every tool across six agent-communication protocols:

---

Live Endpoints

---

Response Contract

Every tool call returns a CanonicalResponse — same shape regardless of protocol:

{
  "agent_id":          "generator_sizing",
  "agent_version":     "1.0",
  "protocol_version":  "1.0",
  "execution_context": "deterministic_calc",

  "status": "success",
  "result": { /* structured data */ },

  "data_freshness": {
    "validated_at":  "2026-03-28",
    "standards_ref": ["NFPA 110:2022"],
    "stale_risk":    "medium"
  },

  "a2a": {
    "task_id":           "uuid-v4",
    "input_tokens_used": 0,
    "credits_consumed":  5,
    "callable_by": ["google-a2a", "openai-agents", "ag-ui", "acp", "agntcy"]
  }
}

Latency

All 50+ individual agents are deterministic TypeScript calculations — no LLM calls, no external I/O.

---

All 50+ Agents

Power & Electrical

Cooling & Mechanical

Network & Connectivity

Site & Construction

Utility & Interconnect

Compliance & Certification

Physical Security

Financial & Operations

---

Architecture

+------------------------------------------------------+
|  MCP Client (Claude Desktop / Claude Code / Cursor)  |
+-------------------------+----------------------------+
                          | SSE (MCP protocol)
                          v
+------------------------------------------------------+
|  orchestrator/  (Node.js on Render)                  |
|                                                      |
|  Auth > Rate Limit > Validation > Credit Gate > Run  |
|                                                      |
|  50+ agents & 8 chains                    |
|  Protocols: MCP, A2A, OpenAI, AG-UI, ACP, AGNTCY   |
+----------+----------------------------+--------------+
           | HTTP (credit check/deduct) | write
           v                            v
+--------------------+       +--------------------+
|  billing-api/      |       |  audit.db          |
|  (Express HTTP)    |       |  (SQLite)          |
|  Stripe webhooks   |       +--------------------+
+--------------------+

---

Pricing

Get a free API key: Sign up

---

Self-Hosting

Prerequisites

• Node.js 18+
• npm 9+

Install & Run

git clone https://github.com/RobotFleet-HQ/security-orchestra
cd security-orchestra

# Build
cd orchestrator && npm install && npm run build && cd ..
cd billing-api  && npm install && npm run build && cd ..

# Configure
cd orchestrator && cp .env.example .env
npm run generate-key myuser free   # generates sk_live_... key

# Run
cd orchestrator && npm start       # Terminal 1
cd billing-api  && npm start       # Terminal 2

Connect Claude Desktop (self-hosted)

{
  "mcpServers": {
    "security-orchestra": {
      "command": "node",
      "args": ["/path/to/security-orchestra/orchestrator/dist/index.js"],
      "env": {
        "ORCHESTRATOR_API_KEY": "sk_live_your_key",
        "BILLING_API_URL": "http://localhost:3001"
      }
    }
  }
}

---

Repository Structure

security-orchestra/
├── orchestrator/              # MCP server — 50+ agents & 8 chains
│   └── src/
│       ├── index.ts           # Entry point, tool registry
│       ├── auth.ts            # API key auth
│       ├── rateLimit.ts       # Sliding-window rate limiter
│       ├── validation.ts      # Input sanitization
│       ├── billing.ts         # Credit check/deduct
│       ├── canonical.ts       # CanonicalResponse shape
│       └── audit.ts           # Audit log
├── billing-api/               # HTTP API — users, credits, Stripe
├── [agent]-agent/             # 50+ individual agent modules
├── smithery.yaml              # Smithery registry config
├── mcp.json                   # MCP registry manifest
└── LICENSE                    # MIT

---

Security

• API key authentication on all tool calls
• Input validation and sanitization on every request
• Sliding-window rate limiting per key and tier
• Audit logging of all tool invocations
• No LLM calls in individual agents — deterministic calculations only
• Credit-gated access prevents abuse

---

License

MIT — RobotFleet HQ