Things move so quickly these days. Why, we could have sworn it was just yesterday when Anthropic open sourced the Model Context Protocol — MCP, to its friends — a standard for connecting AI agents to data, tools and services.
And now it’s time to buy it a smash cake.
Yes, MCP is one year old now, and one of the most popular posts with New Stack readers last week marked the occasion, looking back and forward at what’s ahead for the project. The post by Cody de Arkland, head of developer experience at Sentry, noted that Anthropic’s “lofty” claims for MCP have largely come to fruition.
“In its first year, it’s hard to build a case that says Anthropic overpromised,” de Arkland wrote. “MCP still has room to improve, but the protocol filled a need and has found its audience fast.”
Not that it’s been easy. For instance, in Sentry’s early attempts to build with MCP, he wrote, the code broke a lot, highlighting a need for better observability with the protocol. “[W]e’re absolutely seeing a rise in people wanting a clearer understanding of what’s happening under the hood with their MCP servers. MCP feels like magic until it breaks and you’re digging for why.”
The next phase of MCP will likely see some tool consolidation and standardization of paths. With the announcement Monday of MCP’s donation to the Agentic AI Foundation, a brand spanking-new entity under the Linux Foundation umbrella, there are a lot of milestones ahead.
— Heather Joslyn, editor-in-chief, TNS