Agentic AI
Cloudflare Launches Agent-Focused CLI, Plans to Retire Wrangler
The cloud infrastructure company has unveiled cf, a new command-line tool designed to work seamlessly with AI agents, while phasing out its older Wrangler CLI.
InfoQ
Cloudflare has released an open beta of cf, a fresh command-line interface built from the ground up to streamline interactions between developers and AI agents with Cloudflare's platform. Developed in TypeScript and published under Apache 2.0 or MIT licenses, the tool represents a departure from the company's existing Wrangler CLI, which has accumulated inconsistencies over years of independent team contributions.
Why Build Something New#
Rather than retrofitting Wrangler to support agentic workflows, Cloudflare chose to start afresh. The original Wrangler evolved organically, with different product teams implementing their own conventions—resulting in fragmented terminology (such as "d1 info" versus "hyperdrive get" versus "workflows describe") and limited consistency across nearly 280 command paths. According to senior product manager Matt Taylor and senior systems engineer Samuel Macleod, this created a ceiling for standardization: "Enforcing patterns across teams was virtually impossible."
A new CLI avoided conflicts with how large language models had already learned to interact with Wrangler, freeing Cloudflare to pursue design patterns optimized for agents without maintaining backward compatibility.
Scope and Capabilities#
The cf CLI is orders of magnitude broader than its predecessor. By drawing directly from Cloudflare's OpenAPI schemas—the same definitions powering the company's API documentation and SDK generation—the tool covers more than 3,000 API operations compared to roughly 280 available in Wrangler. This expansion became possible through Forge, Cloudflare's new unified API generation pipeline, which generates CLI commands directly from API specifications.
Key features include JSON output as the default format (enabling AI systems to parse and transform results), command discovery capabilities, and Vite as the integrated development environment with plugin support.
The TypeScript Debate#
The choice to write cf in TypeScript sparked discussion in developer communities, particularly on Hacker News, where skeptics questioned why a CLI would be implemented in an interpreted language rather than a compiled one. Concerns centered on startup latency, portability, and dependency management.
Cloudflare counters that TypeScript provides advantages for both humans and agents: type safety and Language Server Protocol support allow AI systems to better understand and modify configurations with precision. This programmatic approach to configuration—enabled by a TypeScript-based format—also reduces duplication compared to static formats like TOML or JSONC, where semantic information is less accessible to agents.
