Composing APIs and CLIs in the LLM era
Summary
Bradley Walters argues that in the LLM era, enabling models to compose pipelines of shell commands via CLIs is more scalable and cost-efficient than exposing a large set of micro-tools. He demonstrates practical workflows using OpenAPI-driven APIs (via Restish) and OAuth 2.0 flows, discusses token management, and suggests that CLI-based composition can substitute or complement MCP-style tooling for AI-driven automation.