How MCP tools report progress to agents via notifications

On 2026-11-03, the AI section of DEV Community published a tutorial explaining how MCP tools can report progress to an Agent instead of leaving it spinning. The piece opens with the problem: a long-running MCP tool that gives no feedback turns the Agent into a black box. The Agent cannot tell whether the tool is working, has stalled, or should be abandoned, and users staring at a spinner have no better information. The tutorial then walks through the solution the Model Context Protocol provides: progress notifications, which let the tool report incremental progress on the same channel while the call stays open. To use them, the caller passes a progress token along with the request, and the server uses that token to keep sending percentage-based updates or stage markers without ending the call. The client can surface these updates to the user, and the final result is returned separately from the progress notifications themselves. One important caveat the tutorial stresses is that the token is optional: if the client does not supply one, the tool should run normally and simply forgo progress reporting. That is the entire scope of the guidance as published, with no follow-up coverage or changes to the mechanism reported since.