2026-06-26

Rhys Sullivan note on why MCP underdelivered initially and what comes next

useful framing for the current agent-tooling stack wars, less about MCP vs CLI ideology, more about reducing friction while preserving structure and semantics

Original source

Reading Log X Post Historical Backfill Agents AI Infra Developer Tools

What it is: Rhys Sullivan note on why MCP underdelivered initially and what comes next

Gist: argues MCP launched in the GPT-4o / Sonnet 3.5 era before good agent/tooling patterns were understood, so many servers exposed too few capabilities and clients added too much friction; meanwhile bash/CLI-based agents won because they could chain commands, install tools dynamically, and lean on mature shell primitives. His pushback is that this should not end in “just use CLIs”: CLIs hide action semantics and add statefulness, while the better end-state is harnesses that can expose APIs, MCP, CLIs, GraphQL, etc. through one tool catalog

Newsletter angle: crisp explanation of why shell-first agents surged, and why the next layer probably needs to unify API/CLI/MCP rather than pick one