Shopify CEO Tobi Lütke Considers Banning Claude Code

Shopify CEO Tobi Lütke is considering banning Anthropic’s Claude Code from the company until the coding agent changes how it handles project instructions, saying its reliance on CLAUDE.md can create inconsistencies when employees use different AI coding tools.

“I’m thinking about banning Claude Code at Shopify until they change their mind and read AGENTS.md and .agents/skills etc.,” Lütke said in a post on X. “Insisting on only reading CLAUDE.md sometimes leads to split-brain problems when different team members use different tools. Just unnecessary.”

The comments highlight a growing challenge for engineering teams using multiple AI coding agents, where different tools can follow different instruction files and potentially apply different coding conventions or workflows to the same codebase.

AGENTS.md is a Markdown file designed to provide AI coding agents with project-specific instructions, including coding conventions, repository structure, testing commands and other guidance. 

Introduced by OpenAI in August 2025, AGENTS.md is an open standard that gives AI coding agents a consistent set of project-specific instructions to follow across repositories and development tools.

The Linux Foundation describes AGENTS.md as a universal standard for giving AI coding agents a consistent source of project-specific guidance across repositories and toolchains. The foundation said more than 60,000 open-source projects have adopted the format.

Claude Code, meanwhile, has traditionally used CLAUDE.md as its project-level instruction file. This creates a potential split when a repository supports multiple coding agents. If shared guidance is placed in AGENTS.md while Claude Code follows a separate CLAUDE.md, teams may have to maintain multiple sources of instructions or ensure the files stay aligned.

Lütke’s concern is that this can result in different developers getting different behaviour from AI coding tools even when they are working on the same codebase. 

His reference to .agents/skills also points to a preference for keeping reusable agent capabilities and workflows in a shared project-level structure rather than tying them to one coding tool.

The issue comes as Shopify itself continues to build infrastructure for AI agents.

In May, Shopify introduced support for customising AGENTS.md, which it describes as the canonical agent-discovery document for a store. The company said /llms.txt and /llms-full.txt would mirror the content of /agents.md by default.  

As AI coding agents take on more software development tasks, project-level instructions are becoming increasingly important. Files such as AGENTS.md can give these tools consistent guidance on coding practices, project structure and testing requirements.