GeniusFlow Federation

What GeniusFlow Federation does

The GeniusFlow Federation MCP server provides a remote interface for working with evidence-backed claims about named entities. Its public proving ground is CIRCLE, and the associated API exposes entity dossiers, claim verification, packaging, citation, status, and related operations.

The core workflow follows this sequence: entity, claim, grounding, verification, package, and citation. A dossier supplies published claims for an entity. Each claim can be grounded with a source URL and an expected substring taken from that claim. Verification then checks whether the source still contains that literal string. A package can admit or withhold the result, while citation creates a packet over a published claim object.

How it works

GeniusFlow Federation performs a literal string re-check. It does not determine whether the broader world is true, perform semantic fact-checking, or provide a real-time world model. If a circulating figure is missing, the system keeps it as UNKNOWN rather than filling the gap.

The GeniusFlow Federation MCP server is available at https://geniusflow-federation.vercel.app/api/mcp. MCP clients can connect to that URL as a remote server. The listed core tools are gf_dossier, gf_verify, gf_package, gf_cite, and gf_status. Additional tools include gf_report_feed, gf_darshan, gf_return_wire, and gf_openapi; the README identifies the last two experimental adapters.

The same public deployment exposes HTTP routes for direct use: GET /api/dossier?entity=CIRCLE, POST /api/verify, POST /api/package, and POST /api/cite. OpenAPI and machine-readable documentation are also published by the deployment.

Setup and configuration

No account is required, and the README does not specify credentials or environment variables. Configure an MCP client with the remote URL rather than a local filesystem path:

{
  "mcpServers": {
    "geniusflow-federation": {
      "url": "https://geniusflow-federation.vercel.app/api/mcp"
    }
  }
}

The GeniusFlow Federation MCP server runs on a Vercel hobby deployment with best-effort availability and no promised uptime. The runtime engine repository is private, so the documented connection is to the hosted endpoint rather than a locally installed package.

Tools and capabilities

• Retrieve an entity dossier with gf_dossier.
• Re-check a grounded claim with gf_verify.
• Admit or withhold an evidence package with gf_package.
• Create a citation packet with gf_cite.
• Inspect service status with gf_status.
• Access report-feed, OpenAPI, and experimental adapter functions through the additional listed tools.

Use the expected substring from the returned claim as grounding input; the README specifically warns against inventing that value. The repository includes curl and Python examples for dossier, verification, status, packaging, and citation flows.

Limitations and notes

The GeniusFlow Federation MCP server should not be treated as a complete real-world-asset registry, settlement system, persistent memory product, or proof of information advantage. The README also says that the Federation deployment and the project’s Pages site are separate surfaces whose snapshot identities may differ. CIRCLE is inspectable, but that does not establish a complete registry or production service guarantee.

Getting started with this GeniusFlow Federation MCP server

Always refer to the official documentation for the most accurate and up-to-date information.