[hightechmind]

The concurrency finding is the most striking part — MCP forcing sequential API calls because of schema retransmission between calls is a structural penalty that compounds exactly on the multi-step tasks where you'd most want MCP's composition benefits. The ci_failure_investigation example makes it concrete: 15 turns vs 3, 12× cost difference.

Have you looked at whether client-side schema caching could recover that? If the agent doesn't re-transmit tool definitions on every turn, the sequential vs parallel gap should narrow significantly.

[akane8]

Technically MCP tools can be called in parallel as well, but it seems the agent is generally less likely to do so.

In the evaluation I did account for prompt caching, so the multi-turn penalty is already minimized, yet it was still significant enough to make the difference.