MCP is fine if your tool definition is small. If it's something like a sub-agent harness which is used very often, then in fact it's probably more context effic

MCP is fine if your tool definition is small. If it's something like a sub-agent harness which is used very often, then in fact it's probably more context efficient because the tools are already loaded in context and the model doesn't have to spend a few turns deciding to load the skill, thinking about it and then invoking another tool/script to invoke the subagent.