{"version":"2.1.301","anchor":"mcp-task-retry-on-32602-rejected-task-id","canonical_anchor":"mcp-task-retry-on-32602-rejected-task-id","heading":"Long-running MCP tasks retry when the server forgets their ID","tier":"notice","area":"MCP","scope":"individual","heads_up":false,"url":"https:\/\/changelogs.core-directive.com\/v\/2.1.301\/e\/mcp-task-retry-on-32602-rejected-task-id","release_url":"https:\/\/changelogs.core-directive.com\/v\/2.1.301","markdown":"### Long-running MCP tasks retry when the server forgets their ID\n\nIf an MCP server rejects a task's ID with error -32602, Claude Code now waits and asks again instead of failing at once, unless the task has expired\n\n**What**\n\nMCP servers are add-ons that give Claude extra tools. Some of their work runs as a task that takes a while and is checked on by ID. When the server answers that it does not recognise the task's ID (error -32602), Claude Code now waits and asks again, within the time it allows for reconnecting. It stops if the task's own time limit has passed.\n\nBefore, Claude Code gave up straight away and reported the task as expired or unknown to the server.\n\n**Why**\n\nA long-running MCP task can now survive a server restart that briefly loses track of its ID.\n\n- Area: MCP\n- Tier: You'll notice\n- Useful: 2\/5\n- Signal: 2\/5\n- Scope: individual\n- Heads-up: no"}