Contributing Writer · MCP Cost Engineering

Ke Xue

Contributing Writer, Explore Agentic

Ke writes the practitioner side of running MCP at scale: where the attention budget actually goes inside a long session, when re-fetching is cheaper than reuse, and which caching patterns survive a quarter of real workloads. He covers the cost-and-context engineering corner of the MCP pillar.

MCP cost engineeringPrompt cachingContext engineering