Systemic Risk Desk

Click any lens for its full methodology.

The same engines behind this terminal are exposed as Model Context Protocol (MCP) tools, so your AI copilot — Claude Desktop, a custom agent, any MCP client — can query governed systemic risk directly. Every response carries a reproducibility SHA-256 + provenance, so an agent's answer is content-addressed and audit-traceable — the trust gap a regulated institution has when it lets an LLM touch risk numbers. Computed from the public cbsrm methodology only.

cd api
pip install -e '.[mcp]'
cbsrm-mcp            # stdio transport

{
  "mcpServers": {
    "cbsrm-systemic-risk": {
      "command": "python",
      "args": ["-m", "wavervanir_api.mcp_server"]
    }
  }
}

"_governance": {
  "reproducible": true,
  "sha256": "…",
  "as_of": "2026-06-26",
  "note": "Content-addressed and audit-traceable (CBSRM, SR 26-2). Risk measurement, not investment advice."
}

Paste a sanitized broker snapshot (positions only — no account numbers or tokens) and analyze portfolio risk. Anything that looks like a credential is rejected.

Governed, reproducible records of the CBSRM macro-composite pipeline for crisis windows — each content-addressed and verifiable.