{"version":"2.1.319","anchor":"plugin-mcp-servers-can-be-checked-against-the-claudeai-conn","canonical_anchor":"plugin-mcp-servers-can-be-checked-against-the-claudeai-conn","heading":"Plugin MCP servers can be checked against blocked claude.ai connectors","tier":"notice","area":"MCP","scope":"both","heads_up":false,"url":"https:\/\/changelogs.core-directive.com\/v\/2.1.319\/e\/plugin-mcp-servers-can-be-checked-against-the-claudeai-conn","release_url":"https:\/\/changelogs.core-directive.com\/v\/2.1.319","markdown":"### Plugin MCP servers can be checked against blocked claude.ai connectors\n\nClaude Code can check MCP servers supplied by plugins against claude.ai connectors with permission limits, controlled by two remote switches\n\n**What**\n\nPlugins can bring their own MCP servers, which are outside tools Claude Code connects to. Claude Code now builds a list of claude.ai connectors that have a limit on what they may be allowed to do, and checks each plugin server's address against that list. Each match is given a reason:\n\n- address\n\n- host\n\n- opaque\n\n- service\n\n- unreadable\n\nTwo remote settings control this. One turns the check on, and the other only counts matches without acting on them. With neither set, the check is off.\n\n**Why**\n\nOrganisation admins who have restricted a connector may be able to stop plugins from reaching the same service through their own MCP server.\n\n- Area: MCP\n- Tier: You'll notice\n- Useful: 2\/5\n- Signal: 2\/5\n- Scope: both\n- Heads-up: no"}