{"version":"2.1.302","anchor":"parked-mcp-tasks-with-an-active-drive-controller-no-longer-t","canonical_anchor":"parked-mcp-tasks-with-an-active-drive-controller-no-longer-t","heading":"Parked MCP tasks that are being driven are handled more carefully","tier":"notice","area":"MCP","scope":"individual","heads_up":false,"url":"https:\/\/changelogs.core-directive.com\/v\/2.1.302\/e\/parked-mcp-tasks-with-an-active-drive-controller-no-longer-t","release_url":"https:\/\/changelogs.core-directive.com\/v\/2.1.302","markdown":"### Parked MCP tasks that are being driven are handled more carefully\n\nA parked `mcp_task` with an active driver is no longer treated as idle, and a driver error no longer tears the parked task down\n\n**Unclear.** What this changes in what a user sees is not stated.\n\n**What**\n\nAn MCP task (`mcp_task`) is a long-running job run through an MCP server, a tool provider Claude Code connects to. Such a task can be parked, set aside to continue later, and driven, actively worked on.\n\n- When resuming, a parked task now only counts as parked if nothing is currently driving it. Before, any parked task that had not been killed counted.\n\n- If driving a parked task fails and the task belongs to that driver, the driver is stopped quietly and the task is left as it is. Before, the task was always torn down and the error handled.\n\n**Why**\n\nA parked long-running MCP task should survive a passing error while it is being driven, instead of being lost.\n\n- Area: MCP\n- Tier: You'll notice\n- Useful: 2\/5\n- Signal: 2\/5\n- Scope: individual\n- Heads-up: no"}