{"version":"2.1.323","anchor":"mcp-task-polling-defers-to-a-polling-service","canonical_anchor":"mcp-task-polling-defers-to-a-polling-service","heading":"MCP tasks no longer give up on a failed check when a polling service is running","tier":"notice","area":"MCP","scope":"individual","heads_up":false,"url":"https:\/\/changelogs.core-directive.com\/v\/2.1.323\/e\/mcp-task-polling-defers-to-a-polling-service","release_url":"https:\/\/changelogs.core-directive.com\/v\/2.1.323","markdown":"### MCP tasks no longer give up on a failed check when a polling service is running\n\nWhen a polling service is in place, a failed check on an MCP task no longer counts toward giving up on it\n\n**What**\n\nMCP servers are outside programs that give Claude Code extra tools, and some of their tools run as long tasks that Claude Code checks on now and then. When a polling service, a separate service that keeps checking on the task, is in place, a failed check no longer counts toward giving up. Claude Code waits for the next check instead. Before, the first failure moved straight to giving up and checking less often.\n\n**Why**\n\nLong-running MCP tasks are less likely to be abandoned because of one failed check.\n\n- Area: MCP\n- Tier: You'll notice\n- Useful: 1\/5\n- Signal: 2\/5\n- Scope: individual\n- Heads-up: no"}