Throne

Throne is a verification service for MCP servers. It is used to govern the MCP servers that AI agents depend on, and it checks those servers before they enter production.

Its verification flow includes isolated execution, real client behavior, security review, policy decision, and sealed evidence. The page also describes isolated execution as ephemeral and reproducible, and it shows compatibility verdicts and sealed evidence receipts. In an example report, a server is marked with protocol checks passed for Claude Code and Cursor, a clean security review, and a sealed evidence record with a timestamp, scan ID, and verdict.

Throne also presents a public registry and a live evidence layer for public MCP servers from the ecosystem. It says it executes public MCP servers and records what actually happened, and it identifies the resulting public records as independent records rather than endorsements. The page lists example servers such as GitHub MCP Server, Google MCP Toolbox, Grafana MCP, Azure DevOps MCP, Notion MCP, Cloudflare MCP, Stripe Agent Toolkit, Supabase MCP, Sentry MCP, Heroku MCP, HubSpot MCP, Linear MCP, and Figma Developer MCP.

The service is described for teams deciding which MCP servers can be allowed, reviewed, blocked, and put on record. It is presented as a product with live scan, product registry, enterprise resources, and pricing sections, and the page title and metadata describe it as a trust layer for MCP.