Clarification on Claude Code Pro Rate Limits and MCP Usage

Hi Claude team,

Thanks for the great work on Claude Code — it’s been a fantastic tool in my workflow.

I’m currently using the Pro plan, and I understand that there’s a rate limit of approximately 10–40 prompts every 5 hours, depending on complexity.

I have two related questions regarding this:

Does each MCP invocation (e.g. when an agent internally queries a tool via MCP) count toward the Pro plan's rate limit, even if it’s not explicitly initiated by me but triggered as part of an automated reasoning chain?

The documentation recommends that users describe their tasks in more detail up front to save on prompt count. Does this imply that a single long and clear request is more efficient (in terms of usage quota) than breaking it into smaller steps?

Just want to confirm how best to use the tool efficiently under Pro.
Appreciate your clarification!