Strata

Trust and safety layer for AI agents — scores MCP servers for security risk, capability flags, and prompt injection.